Run bybrief-translator
StatusDone
Result5 knowledge gaps
Objective
Work out what is actually being asked: the decision behind the request, the knowledge gaps, and who would have observed each one first-hand.
Questions it set
- KG1Retention economicsasked by client
- KG2The mechanism behind renewal and churninferred
- KG3Head-to-head evaluationsasked by client
- KG4European channel partner economicsasked by client
- KG5Whether the channel is the real route to marketinferred
What it found
- The request is buy-side commercial due diligence: is the revenue sticky, is it defensible, is the route to market profitable enough to keep working.
- The brief asks for current customers, but current customers cannot explain churn. Lost and downsized accounts hold the evidence the retention question needs.
- CrowdStrike and Vectra AI are different kinds of comparator: one a like-for-like detection peer, the other a platform that displaces by consolidation. They need different experts.
- Distributors sit one node away from the margin the client cares about; resellers and MSSPs know the full stack.
Decisions
- Two gaps were added that the client did not ask for: the mechanism behind churn (KG2) and whether the European channel is the real route to market (KG5).
- Three blocking questions were recorded for the client rather than guessed: geography, which CrowdStrike comparison is wanted, and how many calls per population.
How it worked
Read
- 00-brief.md: four sentences naming Darktrace, its ActiveAI platform, two comparators, European channel margins and four populations wanted
Process
Read the brief without researching the market. Resolved the one opaque name, ActiveAI Security Platform, to a product brand rather than an employer.
Labelled every statement EXPLICIT, INFERRED or UNKNOWN so that inferences never silently become client requirements.
Translated the three questions into five knowledge gaps and, for each, described the kind of person who would have seen the answer first-hand.
Tools
Claude WebSearch (lookups only)
Wrote
Run byall-company-names
StatusDone
Result21 names
Objective
Establish every name the target operates under, and which of them people actually list as an employer.
What it found
- 21 names resolved: two live UK operating entities, a German subsidiary, a US federal arm, three acquisitions, five deal-structure shells, and the product brands.
- Five private-equity shell entities employ nobody and were excluded as employers.
- Two live UK entities could each hold payroll; later stages were told to treat both as employers rather than pick one.
- A French subsidiary is claimed but could not be confirmed on the register.
Decisions
- Legacy brands were kept as search vocabulary because older profiles still use them.
How it worked
Read
Process
Worked through the company register, the IPO and take-private structure, subsidiaries, acquisitions and product brands.
Marked each name by type, the period it was in use, and whether anyone would list it as an employer.
Tools
Companies House, German Handelsregister, Claude WebSearch
Wrote
02-company-names/darktrace.md
Run bycompany-researcher
StatusDone
Objective
Research the company and its market to the depth the later stages need: what it sells, the value chain, route to market, the competitive picture, the organisational shape, and what changed recently.
What it found
- The April 2024 relaunch folded older brands into modules priced per coverage area; whether that drove real multi-module expansion could not be settled from public sources.
- The October 2024 take-private ended public disclosure at the year to June 2024, so everything the transaction hinges on is undisclosed.
- Route to market is a mix of direct sales and a two-tier partner programme that is recent and still being built out.
Decisions
- Marked the platform-substance question as unresolved, which later made it a screening question for the independent expert stage.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/darktrace.md
Process
Organised around the six stages that read it rather than as a company profile. Dated every figure and said where the public record stops.
Tools
Claude WebSearch, Claude WebFetch, Annual reports and regulatory filings, Trade press
Wrote
Run byexpert-types-briefs
StatusDone
Result5 of 6 activated
Objective
Decide which of the six expert types can answer a material part of the brief, and write a self-contained brief for each one activated.
What it found
- Five of six types activated: former employees, customers, competitors, channel partners and independent experts. Suppliers were left out as irrelevant to a software vendor’s retention and channel questions.
- Churned and downsized customers were put in scope alongside current ones, correcting the brief.
- Current Darktrace employees were disqualified non-negotiably across every brief.
Decisions
- Writing a brief file is what dispatches a specialist, so this stage sized the entire rest of the run: five briefs, five specialists, three market maps.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/darktrace.md
- 03-company-research.md
Process
Applied the selection rule: default toward recall, but do not activate a type unless that relationship could plausibly answer a material part of the question.
For each activated type, named the population, what they must have done first-hand, the knowledge gaps they own, and the disqualifiers.
Split shared gaps into lettered halves (KG3a, KG3b) so that every part has one owner and two briefs never chase the same person.
Tools
Local reasoning only; no searching
Wrote
04-expert-types-briefs/ex-employees.mdcustomer-experts.mdcompetitor-experts.mddistribution-channels-experts.mdindependent-experts.md
Step 5, market map
Map competitors
Run bycompetitor-finder
StatusDone
Objective
Build the ranked, evidence-backed set of companies that compete with the target, typed by how they compete.
What it found
- “No decision” is the most common competitor: buyers keeping an incumbent rather than switching.
- Vectra AI is the like-for-like peer; CrowdStrike and Microsoft displace by consolidation; email-security specialists displace one module at a time.
Decisions
- Recommended treating an email-security vendor as a first-class competitor population, a call later flagged for the client to confirm.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/darktrace.md
- 03-company-research.md
Process
Searched against a market fingerprint rather than a category label, ran an adversarial pass on its own list, and recorded documented switching stories.
Tools
Claude WebSearch, Exa semantic search, Vendor comparison pages, Analyst reports
Wrote
05-market-map/competitors.md
Step 5, market map
Map customers
Run bycustomer-finder
StatusDone
Objective
Establish which organisations buy from the target, current and former, with each relationship evidenced rather than inferred, and churned accounts hunted deliberately.
What it found
- Three buyers had written down, in legally filed procurement notices, why they would not switch away: accumulated learning, a relearning period, switching costs judged disproportionate.
- Several accounts looked churned because no successor notice was found. Verification later showed four of them had renewed at higher values.
Decisions
- Flagged an unplanned KG5 signal: several accounts bought direct from the vendor with no reseller.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/darktrace.md
- 03-company-research.md
Process
Built the list from UK public procurement notices first, because they name the buyer, the value, the term and often the reasoning, then added vendor-published case studies as weaker evidence.
Tools
Find a Tender, Contracts Finder, Claude WebSearch, Vendor case studies
Wrote
05-market-map/customers.md
Step 5, market map
Map distribution channels
Run bydistribution-channels-finder
StatusDone
Objective
Reconstruct how the product reaches customers: distributors, resellers, MSSPs and integrators, each graded by evidence.
What it found
- One pan-European distributor agreement, announced for seven peripheral markets; the core markets have no distributor above the reseller tier.
- Tiered partners identified in the UK, Ireland, Germany, the Netherlands, Spain and Finland.
Decisions
- Central and Eastern Europe was flagged as unsearched rather than silently omitted.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/darktrace.md
- 03-company-research.md
Process
Started from the vendor’s own partner directory and partner-tier announcements, then worked country by country across Europe.
Tools
Vendor partner directory, Partner press releases, Claude WebFetch, Claude WebSearch
Wrote
05-market-map/distribution-channels.md
Step 6, people search
Find former employees
Run byex-employees-finder
StatusDone
Result12 found
Objective
Find people who used to work at the target, established by employment record and date rather than by job title.
What it found
- Twelve found against a quota of eight to twelve: five renewal or quota carriers, three sales-operations seats, two deployment engineers, one channel lead, one customer-success executive.
- The word “former” in a search does nothing: about 85% of people returned were current employees. Re-wording around the person’s current employer being elsewhere lifted the yield from roughly 2% to 18%.
- Two fabrications were caught in the search layer and recorded so the verifier would not re-import them.
Decisions
- The channel group came back one short of the two asked for, and was left short rather than padded.
How it worked
Read
- 04-expert-types-briefs/ex-employees.md
- 02-company-names/darktrace.md
Process
Ran structured people searches and filtered on the employment end-date field, never on the query wording.
Wrote one evidenced file per person with employment confidence kept separate from relevance.
Tools
Exa people search (structured work history), Claude WebSearch, Public professional profiles
Wrote
06-candidates/ex-employees-finder/ (12 files + run notes)
Step 6, people search
Find customer-side people
Run bycustomer-experts-finder
StatusDone
Result9 found
Objective
Find people who personally operated, evaluated, renewed or cancelled the product at a customer organisation, with the person’s evidence capped by the organisation’s.
What it found
- Nine people at seven organisations, split between current accounts past a post-relaunch renewal and accounts believed to have churned or downsized.
- The only named human in the only confirmed churn story was an outsourced provider, not an employee, and was routed away.
Decisions
- Where the run was short, it was stated rather than backfilled.
How it worked
Read
- 04-expert-types-briefs/customer-experts.md
- 05-market-map/customers.md
Process
Worked every organisation on the map, including the ones that came back empty, and graded product contact as documented, positional or asserted.
Tools
Exa people search, Claude WebSearch, Procurement notices, Vendor case studies
Wrote
06-candidates/customer-experts-finder/ (9 files + run notes)
Step 6, people search
Find competitor staff
Run bycompetitor-experts-finder
StatusDone
Result8 found
Objective
Find people at rival vendors who personally competed against the target, qualified on evidence of the encounter.
What it found
- Eight found, none with a documented encounter, and the reason is structural: the rival’s public versus-Darktrace content is written by executives the brief disqualifies.
- Seven former Darktrace people now at one email-security vendor were found and routed to the former-employees specialist rather than filed here.
How it worked
Read
- 04-expert-types-briefs/competitor-experts.md
- 05-market-map/competitors.md
Process
Screened on employment, then graded the encounter: documented, positional or asserted. Asserted candidates were not filed.
Tools
Exa people search, Claude WebSearch, Vendor pre-sales pages
Wrote
06-candidates/competitor-experts-finder/ (8 files + run notes)
Step 6, people search
Find channel-partner people
Run bydistribution-channels-experts-finder
StatusDone
Result9 found
Objective
Find commercial people at resellers, distributors and MSSPs who personally negotiated or carried the P&L on the vendor line, and can compare it with another vendor they carried.
What it found
- Nine found across six countries; five named and quoted in dated artefacts, four standing opposite a named vendor executive.
- The largest Darktrace-related European contract found, a systems-integration deal of roughly €20m, was identified but not worked for people before the budget ran out.
How it worked
Read
- 04-expert-types-briefs/distribution-channels-experts.md
- 05-market-map/distribution-channels.md
Process
Worked partner by partner across six countries, anchoring each person to a dated artefact naming them opposite the vendor where one existed.
Tools
Exa people search, Claude WebSearch, Partner press releases, Procurement award notices
Wrote
06-candidates/distribution-channels-experts-finder/ (9 files + run notes)
Step 6, people search
Find independent experts
Run byindependent-experts-finder
StatusDone
Result4 found
Objective
Find people who have judged the category repeatedly while selling none of it: former analysts, selection consultants and testers.
What it found
- Four found against a quota of four to five; the fifth slot was left empty rather than filled with an asserted candidate.
- Every named author of every current NDR category evaluation is still employed at their firm, and analysts who leave go to NDR vendors. The ideal profile is systematically unavailable.
How it worked
Read
- 04-expert-types-briefs/independent-experts.md
- 03-company-research.md
Process
Topic-first: worked backwards from published evaluations and surveys to their authors, then checked independence at firm level rather than taking it on description.
Tools
Claude WebSearch, Claude WebFetch, Analyst publications, Conference programmes
Wrote
06-candidates/independent-experts-finder/ (4 files + run notes)
Run byverifier
StatusDone
Result42 tested
Objective
Try to prove each candidate wrong: do the cited sources say what was claimed, do the dates and titles mean what was assumed, is this one real person who belongs to the brief they were filed under.
What it found
- 22 confirmed, 17 downgraded, 1 reclassified, 1 refuted as filed, 1 unresolved. Nothing was misrouted.
- Four UK accounts filed as churned or downsized had in fact renewed at materially higher values. One confirmed churn survived.
- A widely used org-chart site asserted current employment for three people and was wrong every time; it was struck as a source for the rest of the project.
- A distributor’s seven-country footprint claim, inherited by three files, is a press-release claim its own country pages do not corroborate.
Decisions
- Downgrades reduce the strength of evidence, not the wording: a candidate filed as documented and reduced to positional keeps that marker through every later stage.
How it worked
Read
- 06-candidates/ (42 files across five specialists)
- 04-expert-types-briefs/ (the five briefs)
Process
Retrieved the central artefact for each person and read it in full. Tested the eligibility rules of the brief line by line. Wrote a verdict with confidence stated separately from relevance.
Ran in two passes as the later specialists completed; the second pass also wrote a run-level corrections file.
Tools
Claude WebFetch, Claude WebSearch, Find a Tender, Companies House, Public professional profiles
Wrote
07-verification/ (42 verdicts)07-verification/_upstream-corrections.md
Run bymerger-ranker
StatusDone
Result41 ranked, cut at 19
Objective
Combine anyone found twice, rank survivors by the knowledge gap they answer rather than by how good they look individually, and state which parts of the question this run cannot answer.
What it found
- Nobody needed merging: no person was found by two specialists, and the near-collisions were recorded so nobody re-derives them.
- Two people were moved above the cut line on scarcity rather than individual quality: the sole witness to the only confirmed churn, and the only outside judgement on whether the platform relaunch was substantive.
- The channel question flipped from unanswerable to the best-answered of the client’s three once the channel specialist had run.
Decisions
- Cut line after #19, up from #12 in the earlier version, because a previously empty gap acquired witnesses.
- One candidate held out rather than dropped until his employment status is settled.
What this run cannot answer
- No witness to displacement by CrowdStrike platform consolidation; the only surviving evidence is a counter-example where a buyer kept both.
- The vendor side of the channel terms is uncovered: everyone who set them is still employed by the target.
- The only confirmed churn rests on one person whose operational role is unscreened; the best-informed human in that story is an outsourced provider outside the pipeline’s populations.
- Seven of the nineteen people above the cut line are reachable only through an employer inbox or switchboard, and three of those seven sit at partners with a live commercial interest in the company being diligenced.
How it worked
Read
- 06-candidates/
- 07-verification/
- 01-brief-spec.md (the authority)
Process
Weighted the gaps first (retention, then channel economics, then defensibility) and let the people inherit the order. Within a gap: first-hand proximity, evidence strength, period fit, specificity, freedom to speak.
Placed the cut line where every gap the run can cover has at least one witness from a distinct vantage point.
Tools
Local reasoning only
Wrote
Run bycontact-finder
StatusDone
Result21 run, 12 reachable
Objective
Find the professional contact route for one shortlisted person at a time: business email and published professional channels only, every value carrying its source and date.
What it found
- Twenty-one people run: all nineteen above the cut line, one who fell below it on the re-rank and had already been searched, and one blocked at the gate. Twelve have a business email that reaches the person; the other nine have an employer inbox, a switchboard or a professional profile only.
- Before licensed enrichment, exactly one person had a findable personal business email. Enrichment supplied most of the rest, and every dataset value is graded as a dataset’s claim rather than a self-published one.
- Most “found” results are routes to the employer, not to the person. The route type is the actionable fact, not the found/not-found split.
- Three findings came out of this stage that invalidate upstream files: one partner was acquired in November 2025 and the candidate resigned its directorship, so files calling him a current director may be stale; a second partner’s homepage now claims a higher tier with Darktrace than its own Darktrace page does, which undercuts the billing one shortlisted person was ranked on; and a third partner has no separate legal entity in the country the brief requires.
Decisions
- Phone numbers were never sought. One person was blocked at a do-not-contact gate and no details were looked for at all.
- One address published by a candidate herself was recorded as do-not-use rather than as a route, because it sits at the target company.
How it worked
Read
- 08-shortlist.md
- 07-verification/<person>.md
Process
Ran a fixed waterfall per person: the employer’s published routes, the person’s own publications, official registers, professional directories, then one licensed enrichment lookup keyed on the verified profile.
Refused to construct an address from an employer’s naming convention. Recorded any former-employer mailbox as do-not-use, because writing to it would disclose the engagement to its subject.
Tools
RocketReach (licensed enrichment), Claude WebFetch, Claude WebSearch, Employer websites, Find a Tender
Wrote
09-contacts/ (one file per person run)
Step 10, delivery
Publish the case study
Run byvisualiser
StatusDone
Objective
Turn the finished run into one typed data file for this site, leading with what the run cannot answer, and leave every real name, address and phone number behind in the run folder.
What it found
- The content model has no name field for a person at all, and reduces contact to a route type. The boundary is expressed as a type rather than as a rule to remember.
- Compound grades keep their wording rather than being averaged, and confidence stays separate from relevance, as every stage upstream kept them.
Decisions
- Coverage keeps “not covered” and “not searched” apart, and records the one archetype that was never activated as a third state again.
How it worked
Read
- Every file in the run folder
- The site’s content model, which is the contract the data has to satisfy
Process
Rendered, never re-judged: every verdict, rank, grade and correction on this page was made by an earlier stage.
Replaced each person with a role-level descriptor, then generalised the role or the employer wherever the pair would identify one findable individual.
Carried the contact route type only. Nothing from the contact files crosses into the site: not an address, not a number, not a masked fragment.
Tools
Local only, A type-checked build as the verification step
Wrote
One project data file, registered in the site’s project index
Run bybrief-translator
StatusDone
Result4 knowledge gaps, themselves a deliverable
Objective
Work out what is actually being asked: the decision behind the request, the knowledge gaps, and who would have observed each one first-hand.
Questions it set
- KG1Does willingness to pay survive free substitutesinferred
- KG2Who actually controls the purchase, and how sticky is that positioninferred
- KG3How buyers rank the named players, and what makes them switchasked by client
- KG4Is this one market or severalinferred
What it found
- The decision is a pre-LOI category and asset screen by a private equity firm. The real question underneath is whether a category built on read-aloud, word prediction and dictation still supports a paid subscription now that generative speech and OS-native reading are free and general-purpose.
- Two experts is the binding constraint on the whole run: two people must span four topics, so breadth of vantage point beats depth on a single node. A wide candidate pool serves selection, not the deliverable.
- Payer, specifier and user are frequently three different parties in this category, which is why “who is the customer” is not a trivial question here.
- Naming three vendors invites a run built from their alumni. That would answer the fourth topic reasonably and the first three poorly: customer perception belongs to customers and to the intermediaries who compare products for a living.
Decisions
- Flagged a group brand seen only in a search-result title as unconfirmed and refused to treat it as established, handing it to the name-resolution stage rather than assuming it.
- Recorded two blocking questions (whether a specific asset is in play, and which buying regime) and proceeded on stated assumptions rather than stalling.
How it worked
Read
- 00-brief.md: three deliverables (the value chain, four key topics, and the names and positions of two optimally best-placed experts)
Process
Read the brief without researching the market, and labelled every statement EXPLICIT, INFERRED or UNKNOWN so that inferences never silently become client requirements.
Treated the subject as the category rather than any one of the three named players, and wrote the four knowledge gaps as a standing client-facing deliverable rather than as internal scaffolding.
Described, for each gap, the kind of person who would have seen the answer first-hand, and warned that the highest-coverage pair is probably not two ex-vendor people.
Tools
Claude WebSearch (lookups only)
Wrote
Run byall-company-names
StatusDone
Result3 companies, 3 different shapes
Objective
Establish every name each named player operates under, and which of them people actually list as an employer.
What it found
- The unconfirmed group brand was right and bigger than a brand: two companies merged in March 2024, the group rebranded in January 2025 and the register renamed the operating entity in April 2025. Any alumni search on the old name alone misses everyone who has updated a profile since.
- Four companies the brief-stage notes listed as competitors are group-owned. Their alumni belong to the former-employee population, not the competitor population. This is the largest false-positive class in the run.
- The second named player is the opposite shape: one name, one entity, no rename, no parent, no acquisitions, but globally distributed staff with no local entities, so alumni search must not require a national employment record.
- The third named player’s own name finds almost nobody. Its people sit under a hospital that has carried four names, a former North American sales entity, and a UK parent that renamed in 2018.
- The second named player’s blog is an SEO farm of “X versus us” pages naming every rival; any stage researching another vendor had to exclude the domain or read marketing as market evidence.
Decisions
- An ownership contradiction between a press release and an about page was flagged as unresolved rather than summarised into a single claim.
How it worked
Read
Process
Ran three times in parallel, once per named player, working through company registers, renames, subsidiaries, acquisitions and product brands.
Recorded name collisions that would pollute later people searches, and built a leave-date-to-employer-name table where an organisation had renamed several times.
Tools
Companies House, Delaware and Canadian registers, Claude WebSearch, Claude WebFetch
Wrote
02-company-names/ (one file per named player)
Run bycompany-researcher
StatusDone
Resultvalue chain across 5 nodes
Objective
Research the category to the depth the later stages need, and write the value chain as a standalone client deliverable rather than as background.
What it found
- The supplier node holds the live investment question and was unmapped before this run. The market leader’s published sub-processors include four external voice suppliers, and its headline 2025 AI voices are rented from a vendor that sells the same voices to everyone and ships a free consumer reader.
- Two of the three named players share one voice dependency, owned since 2022 by an adjacent assistive-technology competitor.
- The group runs two incompatible AI postures side by side: on-device models with explicitly no third-party API calls in the flagship product, and external LLM vendors named as sub-processors of others.
- The consumer player built its own voice stack, ranked first on a public TTS benchmark, and now sells it as an API, becoming a supplier to the category it competes in.
- Local-language search changed the competitive map. A Flemish scheme funds free reading software against five approved programmes and the market leader is on none of them; Norway pays per individual through a national agency; Denmark buys through municipalities. None of this appeared in English results.
- The UK channel was restructured in February 2024: dozens of one-stop shops became two prime contractors working against a state catalogue. The market leader claims 128,000+ licences through that regime, a position that can vanish with no customer changing their mind.
Decisions
- Recorded, as an honest gap, that no published case could be found of an institution cancelling a paid licence because a free built-in sufficed, and concluded the substitution question was interview-only. The market-map stage later overturned this.
- Named and discounted two AI-generated sources rather than silently dropping them.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/ (three files)
Process
Organised the output around the six stages that read it rather than as a company profile, dated every figure, and said where the public record stops.
Searched in local languages as well as English, on the basis that a purchase regime is described in the language of the country that runs it.
Tools
Claude WebSearch, Claude WebFetch, Published sub-processor lists, Government registers, Trade press
Wrote
Run byexpert-types-briefs
StatusDone
Result6 of 6 activated
Objective
Decide which expert types can answer a material part of the brief, and write a self-contained brief for each one activated.
What it found
- All six types activated, which is not the default. Justified by three companies selling in three motions across seven countries and five buying regimes, plus a brief that asks for the value chain as a deliverable in its own right.
- The supplier brief was the weakest and was activated anyway, because the client asked for suppliers explicitly and the node is unusually answerable: sub-processors are published and suppliers are named in the licence agreement.
- One profile genuinely spans nodes: a commercial person at a voice supplier carrying the accounts of two of the three named players at once.
Decisions
- Writing a brief file is what dispatches a specialist, so this stage sized the entire rest of the run: six briefs, six specialists, four market maps.
- Every brief was written for a category screen rather than a single-asset shortlist, because that superset contains the narrower case and the reverse is not true. The blocking question about which it is was still open.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/
- 03-company-research.md
Process
Split every knowledge gap into lettered parts with exactly one owner each, and named in every brief which gaps it does not own.
Resolved twelve both-sides cases by name with one three-way test worded identically in three briefs: sells or supplies means channel, buys and deploys means customer, neither means independent.
Stated every quota as a ceiling rather than a target, totalling 35-47 candidates.
Tools
Local reasoning only; no searching
Wrote
04-expert-types-briefs/ (six briefs)
Step 5, market map
Map competitors
Run bycompetitor-finder
StatusDone
Result18 ranked
Objective
Build the ranked, evidence-backed set of companies that compete with the named players, typed by how they compete rather than listed flat.
What it found
- Ranked a Nordic listed-parent vendor first rather than the famous American brand, on reasoning: it meets the market leader in four countries, where the famous name is a seven-person brand.
- Two rivals surfaced only on French and Dutch queries, one of them claiming the majority of francophone institutions in Canada, straight at a named player’s home market.
- A national protected-file list does include the market leader’s products, which rules out language as the reason for its absence from the regional approved list. The reason is procurement.
- Germany was searched and came back empty; recorded as an absence rather than an entry.
Decisions
- Ranked a consumer app’s 2024 subscription revolt for the event rather than the company; it is the closest thing to published willingness-to-pay evidence in the category.
- Caught and wrote up a false summary claim that two named vendors share a corporate family, as a worked example rather than a silent deletion.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/
- 03-company-research.md
Process
Typed every row as direct, bundled, substitute, segment, geographic, dual or former, and ran an adversarial pass over its own list.
Applied the carried correction from the name stage, excluding all five group-owned names and recording them as the largest false-positive class.
Tools
Claude WebSearch, Exa semantic search, Local-language queries, Government approved-product lists
Wrote
05-market-map/competitors.md
Step 5, market map
Map customers
Run bycustomer-finder
StatusDone
Result~45 organisations
Objective
Establish which organisations buy from the named players, current and former, with each relationship evidenced rather than inferred and churned accounts hunted deliberately.
What it found
- It overturned the research stage’s key negative. A US university published on its own accessibility blog, in October 2020, that the enterprise agreement would not be renewed because there was no user basis to justify it, naming the free platform-native alternative. It sits on an accessibility subdomain, nowhere a vendor or an analyst looks.
- A 30,000-seat incumbency of the market leader went to open competitive tender at a large US district, with the buyer’s own complaint list and evaluation weights published. The award is public; the winner is not retrievable because the approval document is a scan with no text layer.
- The market leader won a Nordic national tender and then lost it four years later to a rival, displacing its own product: dated, in-window and government-recorded, found only in the local language.
- A named player’s biggest provincial asset ended in 2020 when a central licensing programme closed; it survives on a health catalogue rather than an education one, which is a material downgrade nobody had priced in.
- A vendor-manufactured churn cohort exists right now: two products were retired at the end of 2025, forcing every holding district to reopen a decision on the vendor’s timetable.
- About 34 of ~45 organisations rest on the buyer’s own publishing or a government record. That is the inverse of the usual ratio for this stage. The weakest section, the workplace segment, was flagged as such.
Decisions
- A vendor’s claim of a 30,000-seat deployment at a named university was corrected against the university’s own knowledge base: a capped one-year pilot running alongside a continuing site licence for a rival.
- A candidate organisation whose event falls outside the period window was carried as a stated exception for the ranking stage to accept or strike, rather than quietly included.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/
- 03-company-research.md
Process
Built the list from buyers’ own publishing and government records first, then vendor case studies as weaker evidence, grading each relationship.
Hunted the accounts that stopped rather than the accounts that renewed, in local languages as well as English.
Tools
Claude WebSearch, Claude WebFetch, Procurement portals, University accessibility sites, Local-language government records
Wrote
05-market-map/customers.md
Step 5, market map
Map suppliers
Run bysupplier-finder
StatusDone
Result~40 counterparties
Objective
Identify the companies that supply the named players, what they provide, how material it is, and whether the relationship is one anybody could be interviewed about.
What it found
- The voice-supply layer is consolidating faster than the category it supplies. Three of the four external text-to-speech suppliers changed owner in four years, so the incumbent’s cost of voice is now set by three counterparties all under new ownership needing a return.
- The “no external LLM APIs” posture is a single-product claim, not a group posture: a major LLM vendor is a named sub-processor of four of the group’s products, and one acquired product alone calls four LLM vendors. Both postures are published side by side.
- Speech recognition is entirely bought in from three platform vendors; OCR from three more. The category builds almost none of its own primary inputs.
- Voice talent is the most under-mapped node in the whole chain: no counterparty to any consumer voice deal is published anywhere.
- The consumer player selling voice as an API has exactly one named public API customer, and it is not in education or assistive technology. Strategically true, commercially unproven.
- The market leader contracts with a Swedish legal entity of its voice supplier, not the Belgian one it is usually associated with. A Belgium-only search finds only engineers.
Decisions
- Four sub-scale Nordic hosting providers still serving ten acquired products were recorded as a concrete integration finding rather than as infrastructure trivia.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/
- 03-company-research.md
Process
Found the primary source the research stage lacked (a per-product sub-processor list published as a live document and reviewed in August 2026) and read it product by product rather than relying on a stale single-product page.
Checked each supplier’s ownership history rather than taking its current corporate identity at face value.
Tools
Published sub-processor lists, Claude WebFetch, Company registers, Claude WebSearch
Wrote
05-market-map/suppliers.md
Step 5, market map
Map distribution channels
Run bydistribution-channels-finder
StatusDone
Result8 companies behind a 2-supplier bottleneck
Objective
Reconstruct how the products actually reach buyers (the frameworks, resellers, distributors and approved-supplier schemes in between), with each relationship graded by evidence.
What it found
- It overturned a premise two earlier stages had asserted. Both stated that two of the three named players appear in no UK student-funding material at all. All three are on the state product list, one of them filed under a category that English searches missed.
- Seven products on that same list carry “(Free)” in the name. The free substitute is catalogued by the same body, in the same document, as the paid product. That is substitution evidence sitting in the channel node.
- Who decides the catalogue is neither the funder nor the assessors: applications are administered on behalf of the education department and are accepted from suppliers only. An entry route is published; no exit route exists anywhere.
- The two-supplier bottleneck is a prime-contractor structure of eight companies, not two. One prime was incorporated fourteen weeks before the contract was concluded, and its controlling shareholders are two incumbent one-stop shops.
- A national funder in the Nordics has written the free substitute into its eligibility test: paid software is funded only when the software built into the computer does not meet the need. The sharpest substitution evidence in the run, and it is in the channel node.
- A two-person non-profit is the entire channel for five competing vendors in one region, and you cannot switch product mid-year: switching cost set by the intermediary, not the vendor.
- A price contrast that answers the fourth topic in one line: the market leader sells US districts direct and sole-source at a per-licence price roughly two orders of magnitude below what UK resellers list the same product at.
- The market leader has no partner programme at all, and the consumer player’s channel is genuinely two boxes, said as a finding rather than padded.
Decisions
- Recorded a live dated risk: the UK framework’s initial term ends in November 2026 with no extension or re-tender notice on any of the three public portals.
- Named Germany as the most likely place a further channel exists, never having been covered by any stage.
How it worked
Read
- 01-brief-spec.md
- 02-company-names/
- 03-company-research.md
Process
Pulled the state catalogue’s own spreadsheet, unzipped it and reconstructed the 137-row category-to-product map rather than reading the rendered page.
Worked the corporate structure behind each framework prime through the company register, and read national funding rules in their own languages.
Tools
State catalogue source files, Companies House, Find a Tender / Contracts Finder, Local-language government pages, Claude WebFetch
Wrote
05-market-map/distribution-channels.md
Step 6, people search
Find customer-side people
Run bycustomer-experts-finder
StatusDone
Result10 found
Objective
Find people who personally operated, evaluated, renewed or cancelled a named player’s product at an organisation that bought it, with the person’s evidence capped by the organisation’s.
What it found
- Both sides of the same 30,000-seat US tender were closed: the buyer of record who signed the notice, and the district’s former assistive-technology supervisor.
- The research stage’s “no published cancellation exists” wall came fully down: four institutions stopped, declined or displaced, three of them publishing the reasoning themselves.
- One UK university is stronger than the map showed: it publishes that almost all the market leader’s features are available in other software, and its AT catalogue contains no paid literacy suite at all.
- The workplace segment is structurally unevidenced: all seven named workplace customers rest on the vendor’s own pages, and both workplace files were honestly capped at the organisation’s grade.
Decisions
- Two of the twelve permitted slots were deliberately left unspent and recorded as named worked leads rather than filled.
- Two supersession flags were raised for the verifier: named people who, if shown still in post, would replace the files written in their place.
How it worked
Read
- 04-expert-types-briefs/customer-experts.md
- 05-market-map/customers.md
Process
Worked every organisation on the map including the ones that came back empty, and graded product contact as documented, positional or asserted.
Tools
Exa people search, Claude WebSearch, Procurement portals, Institutional accessibility sites
Wrote
06-candidates/customer-experts/ (10 files + run notes)
Step 6, people search
Find former employees
Run byex-employees-finder
StatusDone
Result9 found
Objective
Find people who used to work at a named player, established by employment record and date rather than by job title.
What it found
- The best file answers three knowledge gaps at once, because the company ran a state-funded catalogue motion and a consumer subscription motion out of one small P&L.
- A conflict was filed rather than hidden: a group-wide revenue leader who is, in his own words, now an advisor to the target’s board, filed with the bar stated in bold rather than quietly dropped.
- Zero candidates at the consumer named player, across five differently-angled searches. Every commercial person of any seniority is still there; only engineering and design people have left.
- The vendor side of the “what does a catalogue position cost to hold” question has no dedicated witness, because the relevant function at the market leader is two people and both are current employees.
Decisions
- An instrument warning was written for the verifier: the people index has merged two of the group’s organisation nodes and renders historical roles of current staff with a null end date, so a null end date does not reliably mean still employed.
How it worked
Read
- 04-expert-types-briefs/ex-employees.md
- 02-company-names/
Process
Ran structured people searches under every historical employer name and filtered on the employment end-date field, never on the query wording.
Tools
Exa people search (structured work history), Companies House, Claude WebSearch
Wrote
06-candidates/ex-employees/ (9 files + run notes)
Step 6, people search
Find channel-partner people
Run bydistribution-channels-experts-finder
StatusDone
Result8 found
Objective
Find commercial people at resellers, distributors and framework contractors who personally negotiated or carried the vendor line, and can compare it with another vendor they carried.
What it found
- Both UK framework primes were covered, not just one, including the candidate who would become the run’s top-ranked person, whose record runs through every seat in the UK chain in order.
- One candidate sat inside a reseller, a framework prime and a catalogue vendor at the same time across the whole transition.
- Six corrections were written back to earlier stages, including two companies listed separately that turn out to be one rebranded business, and an ownership claim about two vendors that are in fact sister brands under a common group.
- A dominant French writing aid, absent from the whole run until this point, is carried alongside a named player’s products by the same distributor.
Decisions
- Three slots were deliberately left unfilled as findings rather than gaps: there is no US channel person to find because the market leader sole-sources direct, no Nordic channel person because the state is the channel, and one distributor’s commercial layer is genuinely invisible with a named search obstacle recorded.
- One candidate was handed back as out of period rather than kept, and two current employees found incidentally were routed to another specialist as ruled-out-but-recorded names.
How it worked
Read
- 04-expert-types-briefs/distribution-channels-experts.md
- 05-market-map/distribution-channels.md
Process
Worked both framework primes rather than one, plus the displaced pre-2024 population, and covered the workplace scheme, Ontario and francophone Canada separately.
Tools
Exa people search, Claude WebSearch, Companies House, Reseller and framework websites
Wrote
06-candidates/distribution-channels-experts/ (8 files + run notes)
Step 6, people search
Find competitor staff
Run bycompetitor-experts-finder
StatusDone
Result7 found, 4 documented
Objective
Find people at rival vendors, including the free and bundled tier, who personally competed against a named player, qualified on evidence of the encounter.
What it found
- Four of the seven have a documented encounter, which is unusual for this population. Two of them build the free and bundled tier that the whole substitution question turns on, a vantage point no other specialist could reach.
- A national displacement in the Nordics was confirmed as government-recorded, naming the product replaced and the date the successor became the only selectable programme.
- The challenger that bid into the large US district tender is documented in the procurement record, and the same-day addendum removed all four sections it contested.
Decisions
- The stage corrected its own dispatch: it had been told the challenger bid against a named 30,000-seat incumbency, and reported that nothing in the RFP, Q&A, addendum or award record named any incumbent. The verifier later reversed this back again by pulling the full Q&A.
- One candidate was written and then cut at the ceiling when a better one surfaced, and the strongest unfiled name was recorded rather than lost.
How it worked
Read
- 04-expert-types-briefs/competitor-experts.md
- 05-market-map/competitors.md
Process
Screened on employment, then graded the encounter as documented, positional or asserted; no asserted candidate was filed.
Tools
Exa people search, Claude WebSearch, Procurement records, Local-language government sources
Wrote
06-candidates/competitor-experts/ (7 files + run notes)
Step 6, people search
Find independent experts
Run byindependent-experts-finder
StatusDone
Result4 found, deliberately under quota
Objective
Find people who have judged this category repeatedly while selling none of it (testers, assessment services and researchers), with independence checked at organisation level.
What it found
- The strongest independent file is government core-funded, sells no software and gives the free substitutes away: the closest thing in the run to someone who has tested the substitution question pupil by pupil for twenty years.
- The substitution researcher the dispatch said might not exist does exist, with fifteen years of empirical work and a published argument for specifying by function rather than by programme, independently the same idea a Canadian province wrote into policy.
- One file’s independence was qualified rather than cleared, and the file says exactly why.
- Zero North American independent experts. Every disinterested voice the run found is European, against a brief whose primary markets include the US and Canada.
Decisions
- Escalated the run’s largest hole: a live government consultation proposing to limit state-funded specialist software to “exceptional circumstances”, in the regime through which the market leader claims 128,000+ licences. Nobody was sourced against it.
- Two candidates were handed to other specialists rather than filed here, because the independence test placed them elsewhere.
How it worked
Read
- 04-expert-types-briefs/independent-experts.md
- 03-company-research.md
Process
Topic-first: worked backwards from published evaluations and research to their authors, then checked independence at organisation level rather than taking it on description.
Tools
Claude WebSearch, Claude WebFetch, Academic publications, Government policy documents
Wrote
06-candidates/independent-experts/ (4 files + run notes)
Step 6, people search
Find supplier-side people
Run bysupplier-experts-finder
StatusDone
Result4 found, first live run of this skill
Objective
Find people on the supplier’s side who personally managed a named player as an account, the ones who know its cost structure, volumes and how it negotiates.
What it found
- The first live run of this skill on any project, and it held: all four employment records confirmed, all four account-ownership claims graded positional rather than documented, and nothing asserted, exactly the honest grading the brief asked for.
- The entity correction from the market map was load-bearing. Searching the supplier’s original country finds only engineers; the account holders sit at a different national entity.
- A finding no earlier stage had: a supplier named under six of the market leader’s products had its UK contracting entity struck off three days before the run, unreported anywhere. Flagged as a client diligence item and not asserted either way.
- The three best non-profile sources for supplier-side commercial people are trade-show exhibitor directories, government AT product registers and conference speaker biographies; all publish the named sales contact where the supplier itself will not.
Decisions
- Two slots were left unfilled for stated structural reasons: the consumer player’s supply side is barred by a contradiction in its own brief, and no counterparty to any consumer voice deal is published anywhere.
- Reported a calibration finding for future runs: a parser reading only the structured work-history field returned empty results twenty-five times and nearly cost the stage its best candidate on the first query.
How it worked
Read
- 04-expert-types-briefs/supplier-experts.md
- 05-market-map/suppliers.md
Process
Screened employment first, then graded account ownership separately, in a population where the relationship is usually contractually private and almost never published.
Tools
Exa people search, Company registers, Trade-show exhibitor directories, Government AT product registers, Conference speaker biographies
Wrote
06-candidates/supplier-experts/ (4 files + run notes)
Run byverifier
StatusDone
Result42 tested, 0 refuted
Objective
Try to prove each candidate wrong: do the cited sources say what was claimed, do the dates and titles mean what was assumed, is this one real person who belongs to the brief they were filed under.
What it found
- 28 confirmed, 13 downgraded, 1 reclassified, 0 refuted. Not one candidate turned out to be the wrong person, a same-name composite or a fabrication, which reflects careful sourcing rather than lenient verification.
- It reversed a correction the orchestrator had passed down. A specialist reported that the US district’s procurement record names no incumbent; the full 41-item Q&A names the incumbent product at item 24, permits it to bid at item 25, and gives 30,000+ deployed student users at items 32-33. The specialist was wrong on two of three limbs.
- It also found a correction that was itself wrong: a specialist had “corrected” a regional approved-programme count from five to four; the fifth is still on the operating body’s live pages, and it is five.
- Both supersession flags resolved. One flagged person is traceable after all; the specialist missed her because she publishes under two surname forms. The other does not supersede anybody: the roster cited sits on pre-production hosts that carry superseded content by construction.
- The most consequential single correction: a candidate filed as a trade association’s executive director had stepped down from that role years earlier. That was one of the file’s two independent sources, so the employment grade dropped. The unlock was a one-line note in another specialist’s run notes about which of two domains actually resolves.
- Namesake hazards in five of twenty-one files in one pass alone, and cached page copies that can be older than the live page even with a maximum-age parameter set.
Decisions
- Delivered explicit rulings so the ranking stage would not have to re-derive them, including one candidate ruled not deliverable on a board advisory role rather than disqualified on employment, a distinction the two should not be allowed to blur.
- The run-level corrections file states the finding the run most needed: the binding constraint here is deliverability, not evidence, and the three cleanly unconflicted candidates were none of them their own folder’s top-ranked file.
How it worked
Read
- 06-candidates/ (42 files across six specialists)
- 04-expert-types-briefs/ (the six briefs)
Process
Retrieved the central artefact for each person and read it in full, in two parallel passes split by folder so no two writers shared a file.
Wrote a verdict per person with confidence stated separately from relevance, plus a run-level corrections file for the findings no individual verdict could carry.
Tools
Claude WebFetch, Claude WebSearch, Companies House, Procurement portals, Wayback Machine via curl
Wrote
07-verification/ (42 verdicts)07-verification/_upstream-corrections.md
Run bymerger-ranker
StatusDone
Result40 ranked, cut at 2, 2 held out
Objective
Assemble the shortlist the client’s question actually needs, rank survivors by the knowledge gap they answer rather than by how good they look individually, and state plainly what this run cannot answer.
What it found
- Zero merges: not one person appears in two folders, under an exact filename or a name variant, and a deliberate name-variant sweep confirmed it. That is the designed outcome of routing rather than an absence of overlap.
- Four clusters nonetheless function as merges for delivery: two principals at one small vendor, two colleagues inside one joint venture, one regional regime seen at two altitudes, and two chairs at one supplier. Deliver at most one from each.
- The pair chosen are the two ends of one transaction: the specifier who decides which product a funded student receives, and the small vendor holding a catalogue position and trying to be that product. Neither has ever met the other’s constraints, which is why the pair is worth more than the sum of two files.
- Ranking by individual quality would have put a different person first and produced a pair with no witness at all to the fourth topic.
- The UK regime is covered from five independent seats: more depth than the client needs on one country, and the reason a coherent pair could be built out of it. The fourth topic is the weakest, and lost its best witness to a conflict rather than to a shortage.
- A structural warning no individual file could show: the UK sector is small enough that two candidates share a trade-body board with two current executives of the target group. A disclosure, not a disqualification.
Decisions
- Three people were moved against their own file’s recommendation, and one against the ranking of the folder it came from, including one filed as the weakest of nine and offered for cutting, who stays.
- Two people were held out rather than ranked, because both turn on a question the client rather than the pipeline has to answer.
- A do-not-quote list closes the shortlist: seven specific figures and framings that must not reach the client, most of them self-reported revenue and retention numbers that are the interview agenda rather than its output.
What this run cannot answer
- The fourth topic, whether this is one market or several, rests on one of the two shortlisted people. The best witness to it was lost to a conflict rather than to a shortage, and a third name would be needed to give it a second source.
- Neither shortlisted expert has run a documented head-to-head between two of the three named players in one buying process. Three other candidates have, and none is in the pair.
- There is no comparison involving the consumer named player anywhere in the run, from any seat. Five differently-angled searches produced no commercial leaver, so the consumer half of the category has no vendor voice.
- Both shortlisted people are UK and both sit inside one funding regime. That coherence is why the pair works at all, and it is also why the client gets no North American buyer, no Nordic voice and no Flemish voice from these two.
- A live government consultation proposing to restrict state-funded specialist software to exceptional circumstances was never searched against. It would remake the regime both shortlisted experts sit inside.
- Two of the five people the contact stage ran have no email route, and for one of them a working self-published address had to be refused because the mailbox now sits inside the group being screened.
- Two of the UK channel candidates sit on the board of the same national trade body as two current executives of the target group. They are not two independent windows on the regime, and the client should be told rather than discover it.
- Who won the 30,000-seat US district re-tender is not established. That the incumbency existed, was named by the district, and went to open tender with a nine-person challenger bidding into it, is.
How it worked
Read
- 06-candidates/
- 07-verification/
- 01-brief-spec.md (the authority)
Process
Ranked the full field of 40 but argued the top two as a pair rather than as the two best individuals, because two people had to span four topics and a pair answering two topics twice would be worse than a jointly complete weaker pair.
Weighted scarcity of vantage point against the gap’s weight in the decision, then let deliverability dominate, because at a cut line of two an expert the client cannot be put in front of is not a shortlist entry however good the file.
Tools
Local reasoning only
Wrote
Run bycontact-finder
StatusDone
Result5 run, 3 with a business email
Objective
Find the professional contact route for one shortlisted person at a time: business email and published professional channels only, every value carrying its source and date.
What it found
- Three of five have a business email that reaches the person; two have a professional profile only. Both not-founds are settled negatives rather than incomplete searches, and one of them is the correct outcome rather than a failure.
- Licensed enrichment was called twice out of five. The free waterfall answered first for the other three: two of them addresses the people had published about themselves, in their own words, in documents that also closed the identity question without ever matching on a name.
- The most important result was a refusal. One candidate has published a working address at the company that was acquired by the group being screened; the domain still resolves, so a question about the target sent there would disclose the engagement to its subject. Recorded in a separate do-not-use table and never in the contact table.
- A second address was refused on the same principle at a different person’s current employer, because a question about the regime would have reached an interested former employer.
- One found address carries an honest currency caveat: the person retired in 2025, the systems that track research staff have dropped him, and the one publishing contact details has not. Recorded as try-first and liable to bounce, with a designed fallback.
- An employer naming convention was present and refused twice. Both recorded addresses are published links that happen to match the convention: corroboration after the fact, never derivation.
Decisions
- Phone numbers were never sought, and a residential correspondence address found on a public register was deliberately not reproduced.
- For an independent adviser with no current employer, a professional profile is the appropriate channel rather than a fallback, recorded as such rather than as a failure.
How it worked
Read
- 08-shortlist.md
- 07-verification/<person>.md
Process
Ran a fixed waterfall per person: the employer’s published routes, the person’s own publications, official registers, professional directories, then one licensed enrichment lookup keyed on the verified profile.
Extended past the cut line of two to rank five, so the client has reachable fallbacks; a two-name shortlist with no contactable substitute is fragile.
Tools
RocketReach (licensed enrichment), Claude WebFetch, Claude WebSearch, Companies House, Wayback Machine via curl
Wrote
09-contacts/ (one file per person run)
Step 10, delivery
Publish the case study
Run byvisualiser
StatusDone
Objective
Turn the finished run into one typed data file for this site, leading with what the run cannot answer, and leave every real name, address and phone number behind in the run folder.
What it found
- Anonymisation was harder here than on a single-company run. Where a company is small enough to be the identifier, the company was generalised to its kind rather than the role widened, the opposite of the usual preference.
- The do-not-quote list from the ranking stage was applied as a hard filter: no self-reported revenue or retention figure from any candidate file appears on this page.
Decisions
- Coverage keeps “not covered” and “not searched” apart. Six rows are searched-and-nobody-exists, which closes those lines of enquiry; one row is nobody-looked, and reopens for the price of one specialist run.
- The one open item that is a retrieval problem rather than a coverage problem was kept out of the coverage map and recorded under next steps instead.
How it worked
Read
- Every file in the run folder
- The site’s content model, which is the contract the data has to satisfy
Process
Rendered, never re-judged: every verdict, rank, grade and correction on this page was made by an earlier stage.
Replaced each person with a role-level descriptor, then generalised the role or the employer wherever the two together would name one findable individual, which on this run they repeatedly would, because several of the organisations involved have fewer than ten people.
Carried the contact route type only. Nothing from the contact files crosses into the site: not an address, not a number, not a masked fragment.
Tools
Local only, A type-checked build as the verification step
Wrote
One project data file, registered in the site’s project index
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile not found
Role
Global VP, Customer Success, 2022-2025; built the function over the six years before
Darktrace
Why they fit
Was accountable for the retention metric the client is buying and has published net-retention and gross-churn figures on his own firm’s site. The only person in the run who has put a number on KG1 in public, and the only one who can be asked to date and define his own.
Ranked 2, shortlisted
Senior security lead at a UK housing association
CustomerAnswers KG2, KG1a, KG3b, KG5a
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile not found
Confidence and relevance are graded separately.
Confidence: high on the organisation, medium on the person
Role
In the security function; ran a multi-vendor detection evaluation
Current customer; £1.1m three-year direct award, June 2026
Why they fit
Best organisation-level evidence in the project: a legally filed procurement notice in which the buyer writes down why it will not switch. Four named modules, accumulated AI learning, a relearning period, switching costs judged disproportionate. Runs CrowdStrike Falcon alongside Darktrace, the run’s only KG3b evidence. Four gaps in one call.
Check before booking a call
Was Darktrace one of the six solutions he evaluated?
Confidence Low-med
Relevance High
Verification Downgraded
Contact Business email found Email found, profile not found
Confidence and relevance are graded separately.
Confidence: high on the churn, low-medium on the person Relevance: high or near-zero, decided by the screen
Role
Ran the IT estate when the account left Darktrace
Australian car-park operator; replaced Darktrace at ~35% below the renewal quote
Why they fit
Moved above the cut line as the sole witness to an otherwise empty gap. The account is the only confirmed churn left standing: 450+ IPs, seven stated reasons for leaving, three of which the candidate file did not carry. If the outsourced provider ran it, he is a corroborating voice on the cost case rather than the witness.
Check before booking a call
Did he, or the outsourced provider, operate the console and hold the renewal conversation?
Changed during checking
The only person quoted in the churn case study is the account’s outsourced provider, not him.
Confidence High
Relevance High
Verification Confirmed
Contact Employer inbox or switchboard only Email not found, profile not found
Confidence and relevance are graded separately.
Confidence: strongest-evidenced file in the channel stage
Role
Senior alliances role, EMEA, spanning the distribution agreement
Pan-European IT distributor
Why they fit
The only distributor-side witness in Europe, against a spec that says both channel nodes are needed. Documented across ten dated artefacts, with an individual-level comparison against two other vendors that survived a direct falsification test. Authoritative on what Darktrace conceded for shelf space; untested on what the line earns.
Check before booking a call
Did he set the discount and rebate thresholds, or sign what somebody reporting to him negotiated?
Confidence High
Relevance High
Verification Confirmed
Contact Employer inbox or switchboard only Email not found, profile found
Confidence and relevance are graded separately.
Confidence: two independent third-party sources
Role
A decade carrying key accounts, to November 2025
Darktrace
Why they fit
The only person who sold on both sides of the 2023 commission change and through the first post-ActiveAI renewal cycle from a quota seat. Can judge module expansion on the same accounts before and after.
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile not found
Confidence and relevance are graded separately.
Confidence: upgraded on her own dated farewell post
Role
Eight years on quota, five as a regional team lead; left August 2026
Darktrace
Why they fit
The only person sourced who was inside for the entire undisclosed period: both post-ActiveAI renewal cycles, the full restructuring and the early-2026 CEO change. Cleanest current employer in the run.
Check before booking a call
Which sub-region, and how many sellers reported to her?
Changed during checking
A self-published address on her own profile turns out to be a former-employer mailbox at the target and was recorded as do-not-use rather than as a route; writing to it would disclose the engagement to its subject.
Confidence Medium
Relevance High
Verification Downgraded
Contact Employer inbox or switchboard only Email not found, profile not found
Confidence and relevance are graded separately.
Confidence: high on the facts, medium on what he personally said
Role
Security business lead; cleared the German threshold in twelve months from a standing start
German IT system house, Elite-tier partner 2025-26
Why they fit
KG4’s missing number. The only person identified in Europe who can describe Darktrace’s top partner tier from inside it, in a market with no distributor above him. The sharper question is why the tier appears not to have been renewed while rival vendor tiers were being won in the same window.
Check before booking a call
Does the Darktrace contract sit with him, with group procurement, or with a regional managing director?
Changed during checking
A quoted phrase about favourable terms is the partner’s unattributed website copy, not his words; the tier is a 2025-26 status, not a current one.
Confidence Medium
Relevance High
Verification Downgraded
Contact Employer inbox or switchboard only Email not found, profile not found
Confidence and relevance are graded separately.
Confidence: identity and independence high; the Darktrace track record unevidenced Relevance: high or near-zero, decided by the screen
Role
Runs a priced, methodologically specific testing offer against real attack traffic
Independent digital forensics firm
Why they fit
The only candidate anywhere in the run who has instrumented Darktrace against real attack traffic, and structurally the most disinterested party: paid by the buyer, sells no replacement. But what is documented is a published service, not a history of engagements. If the practice began in 2026, KG1’s product-substance question has no evidenced answer in this run.
Check before booking a call
How many Darktrace validations has he delivered, and does the practice predate April 2024?
Changed during checking
The documented grade attaches to a published, priced offer rather than to delivered engagements; all four write-ups carry the same 2026 date.
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile found
Role
Director spanning the IPO and the take-private from one seat
UK managed security service provider running four platforms in its SOC
Why they fit
Promoted relative to where his own specialist placed him: verification found the firm’s SOC runs Darktrace alongside Microsoft Defender, the named platform-consolidation competitor. The only person who can answer the time-series question about how channel terms changed across the IPO and the take-private.
Check before booking a call
How large is the firm’s Darktrace book? It bounds the numbers, not the judgement.
Changed during checking
Filed with the stage’s weakest comparison grade; verification found a documented four-platform SOC and raised it. The contact stage then found the firm was acquired in November 2025 and he resigned the statutory directorship the day before completion, so the upstream files describing him as a current director may be out of date, and an employer mailbox is now one the acquirer controls.
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile not found
Confidence and relevance are graded separately.
Confidence: high on the organisation, medium on the person
Role
Inherited a deployment he did not choose, then renewed it at a ~56% uplift
Current customer; contract grew from £567k (2021) to £885k (2025)
Why they fit
The sharpest available test of the switching-cost moat: his employer filed a four-year learning process as its reason not to compete the contract. Ask what the uplift bought. Do not quote the notice’s words back to him as his own.
Check before booking a call
Does the platform sit under him, or under one of two named colleagues?
Confidence Med-high
Relevance High
Verification Confirmed
Contact Business email found Email found, profile not found
Confidence and relevance are graded separately.
Confidence: triply self-corroborated; no vendor-side source
Role
CCO January 2023 to July 2025, mapping almost exactly onto the brief’s window
Dutch reseller: Elite with three vendors, Premier with Darktrace
Why they fit
The sharpest single KG4 question in the run, and not answerable from any document: you are Elite with three vendors and Premier with Darktrace; what would Elite have cost you, and why was it not worth it? He ranked three vendor programmes and put Darktrace below all of them.
Check before booking a call
Does his current role still see vendor terms, or is his commercial knowledge thirteen months stale?
Changed during checking
The Premier tier is declared by the partner and uncorroborated externally; Darktrace sits below three vendors, not two. The contact stage then found the partner’s homepage now claims the top tier with Darktrace as well while its own Darktrace page still says the tier below, which undercuts the billing this person was shortlisted on; the screen should ask when, and whether, that changed.
Ranked 12, shortlisted
Former NDR pre-sales lead at two rival vendors
CompetitorAnswers KG3a
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile found
Confidence and relevance are graded separately.
Confidence: employer-published
Role
Built and ran one rival’s proof-of-value programme, then two years of NDR pre-sales at another
Rival network-detection vendors (both former)
Why they fit
A clean alumnus of both relevant employers, so nothing constrains him. Twenty-three in-window months of NDR pre-sales, plus the losing vendor’s account of why the category consolidated, available from nobody else.
Check before booking a call
Was Darktrace across the table in a proof-of-value he ran?
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile not found
Confidence and relevance are graded separately.
Confidence: high on the transaction and the role, medium on whether he touches Darktrace Relevance: high or near-zero, decided by the screen
Role
Vendor alliances; nine prior years at a distributor
Irish reseller; €3.5m five-year Darktrace reseller framework, publicly awarded
Why they fit
The cleanest documented instance in the run of channel revenue that is invoiced rather than generated, which is what KG5 actually asks. The same buyer had given the reseller a much larger award four months earlier, which makes customer-originated fulfilment the likeliest explanation.
Check before booking a call
Who inside the reseller owns the Darktrace vendor relationship, and is it him? If the answer is nobody, formally, that is the finding.
Ranked 14, shortlisted
Senior technical pre-sales at a rival NDR vendor, UK
CompetitorAnswers KG3a
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile found
Role
Five years in the UK pre-sales seat, across the relaunch and the take-private
Rival network-detection vendor; now at an endpoint platform vendor
Why they fit
The KG3a witness the spec describes almost word for word: five years selling the like-for-like peer in Darktrace’s home market, twenty-seven years in one territory. Bill him as KG3a only; his later endpoint-vendor seat is a territory role, not a consolidation-business-case seat.
Check before booking a call
Can he name a UK account where both vendors were in the proof-of-value?
Confidence High
Relevance High
Verification Confirmed
Contact Employer inbox or switchboard only Email not found, profile not found
Confidence and relevance are graded separately.
Confidence: upgraded to confirmed
Role
Nine and a half years: five installing on customer sites across EMEA, four owning the technical side of the partner channel
Darktrace
Why they fit
The only substantive vendor-side KG5 witness. Do not brief him on KG4: he held the technical partner seat, not the commercial one. Two namesake and attribution traps were closed in verification.
Check before booking a call
Was his later employer a Darktrace customer while he ran its security? If so he has been on both sides of the product.
Changed during checking
Partner-programme material a search attaches to him belongs to a current executive; he is also not the similarly named public figure.
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile found
Confidence and relevance are graded separately.
Confidence: upgraded; vendor-issued certification
Role
Nearly five years of solutions engineering to August 2026; vendor-certified analyst
Darktrace
Why they fit
His incident-recovery seat spans the launch and the first renewal cycle from the post-sale side: he can say whether the modules customers expanded onto were configured and used or merely licensed, the operational half of KG1’s central question. Not a source for discounting or renewal pricing.
Check before booking a call
How much of the estate did the incident-recovery beat actually touch: a handful of escalations, or a systematic post-sale beat?
Confidence Med-high
Relevance High
Verification Confirmed
Contact Employer inbox or switchboard only Email not found, profile not found
Confidence and relevance are graded separately.
Confidence: employment single-sourced Relevance: high or near-zero, decided by the screen
Role
July 2024 to October 2025, across the take-private and the first post-relaunch cycle
Darktrace
Why they fit
The only seat that sees churn and discounting across the whole base. Best-placed person to say whether the reported net retention figure was constructed generously.
Check before booking a call
Was retention and churn reporting actually inside his remit, or only pipeline and quota?
Changed during checking
An org-chart site listed him as current staff; struck. His own record gives the title and the dates.
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile not found
Confidence and relevance are graded separately.
Confidence: quote confirmed verbatim, counterparty confirmed
Role
Owns the managed-service product line; quoted opposite Darktrace’s channel VP in a dated release
Irish managed security service provider
Why they fit
The only candidate who answers what the brief singles out as the MSSP’s distinctive contribution: the operating cost of running the platform for somebody else, and whether the margin survives it. The unpaid proof-of-value, the renewal-registration carve-out and the certification cost are things he has modelled and nobody has published.
Check before booking a call
Does he hold the vendor contract, or only the delivery P&L?
Confidence Med-high
Relevance High
Verification Confirmed
Contact Employer inbox or switchboard only Email not found, profile found
Role
Leads delivery across three detection platforms
Spanish managed security service provider
Why they fit
Paired with the Irish MSSP director and neither substitutes for the other: one sees product and service margin on one vendor, this one sees delivery cost across three. The firm’s own site documents skilled configuration work on both Darktrace and CrowdStrike, as close as a public source gets to proving the comparison is lived.
Check before booking a call
Does he see vendor margin at all, or only delivery cost? And is what he describes Spanish or Latin American?
Changed during checking
The contact stage found the Spanish operation has no separate legal entity: the certified body is the listed parent group, and the entities carrying the security brand are the Latin American ones. That sharpens the second half of his screen rather than answering it, and it means a register search on the brand in Spain lands on a renamed shell.
Ranked 20, below the cut line
Vendor-management counterparty at the largest German system house
Channel partnerAnswers KG4a
Confidence Medium
Relevance High
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: the 2024 record triple-sourced; current employment unestablished Relevance: high or near-zero, decided by the screen
Role
Security vendor management in the central procurement entity
Largest German IT system house
Why they fit
The named vendor-management counterparty for the largest German system house to take Darktrace on. A rival vendor’s partner award independently places him inside the firm’s security organisation.
Check before booking a call
Two hard gates: is he still at the firm, and did the firm transact?
Changed during checking
The “financial incentives” hook does not survive: two of the file’s own sources gloss the phrase as the published partner-programme benefits.
Ranked 21, below the cut line
Former industry analyst who covered network security 2022-2024
Independent expertAnswers KG3a, KG1a
Confidence Med-high
Relevance Med-high
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Ran a network-security research practice; now at a research and advisory firm
Industry analyst firm (former); advisory firm (current)
Why they fit
The only person whose vantage point across 2022-2024 was the whole network security market seen from inside vendor briefings, taking all three vendors’ briefings in the same period. Positional and unproven on Darktrace specifically: he ran the practice, he is not shown to have scored the product.
Check before booking a call
Will his current firm permit a paid expert engagement on a market it covers?
Ranked 22, below the cut line
Security business-unit manager at a Nordic distributor branch
Channel partnerAnswers KG5a
Confidence High
Relevance Medium
Verification Reclassified
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on what the sources show, medium on the interpretation
Role
Twelve years onboarding and retiring vendor lines; nine vendors in his portfolio
Nordic branch of the only European Darktrace distributor
Why they fit
Reclassified: the branch’s own line card does not carry Darktrace eight months after the distribution agreement. Do not use him as a Darktrace-line witness and do not drop him. Why not, and what does that say about what the agreement is worth, is a direct KG5 answer the run has from nobody else.
Check before booking a call
Has the branch in fact begun carrying Darktrace despite its own pages?
Changed during checking
Sourced as a second distributor-side Darktrace-line witness; the distributor’s own line card and team assignments say he does not carry it.
Confidence High
Relevance Med-high
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on the organisation, medium on the person
Role
Signed a renewal extended to two new sites, December 2025
Current customer, bought direct
Why they fit
A documented renewal and expansion twenty months after the relaunch. The contract carries an option to add or remove parts of the network by variation, a real caution about reading any published value as evidence of stability.
Check before booking a call
Was the renewal consultant-led? The organisation has engaged a senior information-security consultant since late 2023.
Ranked 24, below the cut line
Independent SOC consultant running a ten-year SOC survey
Independent expertAnswers KG3a
Confidence High
Relevance Med-high
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: the change is to the independence grade
Role
Consultant and survey author; SANS-affiliated
Independent consultancy
Why they fit
What he uniquely brings is a denominator: a ten-year census of what SOCs actually deployed and kept. His finding that most analysts treat endpoint alerts as their primary trigger is the mechanism by which a consolidation pitch wins without a detection contest.
Check before booking a call
Has he taken paid work from, or presented for, any of the three vendors inside three years? If so he should be withdrawn.
Changed during checking
Independence moved from cleared to qualified on a 2021 webinar co-presented with, and hosted by, one of the three vendors.
Ranked 25, below the cut line
Former revenue-operations lead, second renewal cycle
Former employeeAnswers KG1a, KG1b
Confidence Med-high
Relevance Med-high
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: employment single-sourced
Role
Revenue operations through the second post-ActiveAI renewal cycle, to April 2026
Darktrace
Why they fit
The second post-ActiveAI renewal cycle, where repackaging-versus-real-expansion actually settles, from the revenue-operations seat. Together with the earlier RevOps director he covers July 2024 to April 2026 continuously.
Check before booking a call
Was retention reporting inside the remit, or only systems?
Ranked 26, below the cut line
Former sales director, EMEA, based in Munich
Former employeeAnswers KG1a, KG2
Confidence High
Relevance Med-high
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: two first-party dated posts
Role
Account executive to sales director; left June 2026
Darktrace
Why they fit
The run’s only DACH witness. He sold before the 2023 commission change and managed sellers through it, a different and arguably better vantage point on KG1 than the file claimed.
Check before booking a call
Was the EMEA title a full EMEA remit, or DACH-anchored with a regional label?
Changed during checking
Left in June 2026, not August; and did not carry a personal quota on both sides of the commission change.
Ranked 27, below the cut line
Hands-on security lead at a UK housing group with a growing account
CustomerAnswers KG2, KG1a
Confidence Medium
Relevance Med-high
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: product contact top-of-band positional
Role
Credited twice by his own IT head with running the programme
Current customer; successor term signed May 2026 at a higher value
Why they fit
Hands-on at a confirmed multi-year account, publicly named by the vendor’s regional account director. He answers why an account grew, not why one shrank.
Check before booking a call
Did he personally sit with the console, or direct others who did?
Changed during checking
Filed as a downsize witness; the account’s successor term is larger than the one it replaced.
Confidence High
Relevance Med-high
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: upgraded; two self-published domains
Role
Carried the same product into both weighted geographies
Rival network-detection vendor
Why they fit
The only person who carried the same competing product in both Europe and the US. Bill the cross-market comparison as asymmetric: seven in-window months in EMEA against sixteen in the US.
Check before booking a call
Did the early-2024 senior EMEA role carry a live book, or was it a transition before the US move?
Confidence High
Relevance Medium
Verification Refuted as filed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on the organisation, low on the person
Role
Public-sector IT lead
Current customer; renewal and expansion at roughly 2.6× the prior value, March 2024
Why they fit
A named public-sector expansion witness at exactly the transition: the award, described as a renewal and expansion, landed twelve days before the ActiveAI launch. Civil-service reachability is a real constraint.
Check before booking a call
Did he or a named colleague run it?
Changed during checking
The refutation is of the account premise (the body did not churn), not the person. Re-placed against the gap he now answers.
Ranked 30, below the cut line
Former strategic renewal manager, EMEA
Former employeeAnswers KG1a, KG1b
Confidence High
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on employment; borderline positional on relevance
Role
Fourteen months across the launch, the first renewal cycle and the take-private; eleven prior years in renewal sales at a large software vendor
Darktrace
Why they fit
A renewal-desk calibration nobody else has, straight out of a decade running renewals elsewhere. Do not brief him on channel: the distribution experience belongs to the prior employer.
Check before booking a call
Did the EMEA title correspond to an EMEA book, or a UK and Ireland book with a regional label?
Changed during checking
A search summary credited him with fifteen years at Darktrace; the tenure is fourteen months and the fifteen years belong to a prior employer.
Ranked 31, below the cut line
Former enterprise account executive, DACH
Former employeeAnswers KG1a
Confidence Med-high
Relevance Medium
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: employment single-sourced Relevance: medium or near-zero, decided by the screen
Role
Twenty-eight months sitting almost entirely inside the priority window
Darktrace
Why they fit
The seller’s view against the manager’s, as the junior half of a DACH pair. Best availability in the run: no employer since late 2025 and actively looking.
Check before booking a call
Did his enterprise book include renewals, or only new business? If it was a hunting seat his KG1 value collapses.
Ranked 32, below the cut line
Former head of channel, nine months
Former employeeAnswers KG4c, KG5b
Confidence Medium
Relevance Medium
Verification Downgraded
Contact Professional profile is the only channel Email not found, profile found
Confidence and relevance are graded separately.
Confidence: employment upgraded to strong
Role
Ran the channel function for nine months; twenty years of channel at two large network vendors before
Darktrace
Why they fit
Not the KG4 witness and must not be presented as one: she inherited and administered the partner terms, she did not write them. What is genuinely valuable is the benchmark: whether Darktrace’s partner economics were competitive or thin against two other vendors’ programmes.
Check before booking a call
Was the engagement employment or a senior interim, given the concurrent consultancy? And who did she report to?
Changed during checking
Was above the cut line only as the closest person to an empty KG4; fell eleven places when the channel specialist ran, without any change to her evidence.
Confidence Medium
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: positional
Role
Commercial counterpart to the hands-on security lead at the same account
Current customer
Why they fit
Useful as the commercial counterpart at a growing account: what the May 2026 uplift bought. Expect him to defer technical detail; his own post says he was supporting.
Check before booking a call
Did he, procurement, or a named colleague lead the negotiation?
Changed during checking
The file’s whole case was a halving of the account that did not happen.
Confidence High
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on employment, medium on the argument
Role
In the pre-sales seat across the whole window; employer documents the function publicly
Rival network-detection vendor (current)
Why they fit
The only Southern Europe representation in the run, in the right seat. A current employee close to being the vendor’s technical spokesperson in her market: the candidate most likely to give positioning instead of a deal.
Check before booking a call
Is her remit France or Southern Europe, and does she still personally sit in proofs-of-value?
Ranked 35, below the cut line
Former security-operations analyst, now independent commentator
Independent expertAnswers KG3a
Confidence Medium
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: independence facts established and unflattering Relevance: medium or near-zero, decided by the screen
Role
Former analyst-firm SecOps coverage; now runs a small research and advisory practice
Industry analyst firm (former)
Why they fit
Not disqualified, but everything else pushes one way: his analyst category was security operations rather than NDR, and his current output is mostly daily commentary. What he offers is decision context: what a SOC already owns and what a new detection tool must displace to get bought.
Check before booking a call
How many NDR vendor briefings did he take, and did he ever write on the NDR market?
Confidence Low-med
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Role
One of a small IT team alongside an outsourced virtual CISO
Customer, status unknown; pre-ActiveAI
Why they fit
Not in the churned half: the original award was below the publication threshold, so a missing successor notice proves little. What makes him worth a call is the shape: a high per-device cost on a small estate, then a move of budget from detection to posture management.
Check before booking a call
Did the organisation renew? And was he the person who ran it, rather than a colleague or the outsourced vCISO?
Confidence High
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on employment, low on the encounter Relevance: medium or near-zero, decided by the screen
Role
Inside the competitive-positioning team; former Darktrace seller 2018-2021
Endpoint platform vendor (current)
Why they fit
Can describe what the endpoint vendor’s field is told to argue. He does not run the competitive-intelligence function, and nothing evidences his remit covering network detection or Darktrace.
Check before booking a call
Has his competitive-intelligence remit ever covered NDR or Darktrace? If no, the brief does not admit him at all.
Changed during checking
Two more senior people sit above him in the function; the file implied he ran it.
Confidence Med-high
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: scope undecided
Role
Three years running technical evaluations, two carrying a mid-market quota
Email-security vendor (current)
Why they fit
Can describe an email-module displacement technically and commercially in one conversation. The only candidate in the run outside the brief’s named populations, admitted on the market map’s recommendation; that decision should be put to the client.
Check before booking a call
Does the client want this vendor in scope at all?
Changed during checking
Not a career email-security engineer, and mid-market rather than enterprise.
Confidence Medium
Relevance Low-med
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: medium on employment, low on identity
Role
Ten years of enterprise security pre-sales in one territory
Endpoint platform vendor (current)
Why they fit
The territory is genuinely his. Everything else is self-reported: the consolidation credential is his own headline, and his surname is inferred from a profile slug rather than established.
Check before booking a call
Has a network or email detection product ever been retired into an endpoint-platform deal in his accounts?
Changed during checking
Identity rests on a profile slug; the profile renders only an initial for the surname.
Confidence Low-med
Relevance Low-med
Verification Downgraded
Contact Below the cut line; contact search not run
Role
Owns the network the product sits on
Current customer
Why they fit
Real, in post, correctly routed, and his qualifying evidence reduces to role plus tenure plus a shared mention in a vendor account director’s post. Third file from one account.
Check before booking a call
Did he personally work with the product at all, or only own the network under it? Ask before offering him.
Changed during checking
The file’s most attractive sentences about sensors and baselining were unsourced hypotheses about what he would know.
Confidence High
Relevance Low
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on employment, medium-low on relevance
Role
Six NDR years on a US federal team with a SIEM specialism; two months in the current seat
Endpoint platform vendor (current)
Why they fit
His NDR years turn out to be US federal years. He occupies the endpoint-vendor allocation on a seat taken up two months before this run; if a slot must be freed, this is it.
Check before booking a call
Did Darktrace ever appear in his deals, federal or commercial? A no removes him.
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile found
Confidence and relevance are graded separately.
Relevance: the shortlist grades this “very high”, the anchor of the pair
Role
Eight years writing named products into funded needs-assessment reports, then subject-matter expert on the prime contractor’s framework bid, then technical management of its live service, 2023-2025; now on the distribution side
One of the two prime contractors on the UK Disabled Students’ Allowance framework; since 2025 an assistive-technology distributor
Why they fit
The only candidate in the run whose first-hand test passes on a documented act rather than a commercial title. For eight years the job was choosing named products from the state catalogue (which carries all three named players, their rivals and seven entries with “(Free)” in the name) for individual named learners, hundreds of times. That is not proximity to the purchase decision; it is the purchase decision. Then the framework prime’s bid, then its live technical service across the February 2024 go-live: the specifier who sells nothing, the framework supplier’s employee and a distributor, three economic interests in one transaction held consecutively by one person.
Check before booking a call
What the current employer actually sells, and whether the current role creates a vendor interest.
Changed during checking
Ranked first overall despite not being the top-ranked file in the folder as originally sourced. Deliverability moved this candidate up: the framework-prime employment ended in 2025, so no live commercial position inside the framework is being protected.
Confidence High
Relevance High
Verification Confirmed
Contact Professional profile is the only channel Email not found, profile found
Confidence and relevance are graded separately.
Relevance: the shortlist grades this “very high”, the only deliverable witness to the pair’s fourth topic
Role
Held the commercial side of a sub-ten-person software company that ran a state-funded catalogue motion and a worldwide consumer subscription out of one P&L; stayed through the acquisition and left the following year
A small vendor holding a position on the UK student-funding catalogue, acquired into the multi-brand group in 2024
Why they fit
The client’s KG4 question (institutional entitlement purchasing against consumer app-store subscription) run by one person, in one company, at the same time, across the exact framework transition, and then sold into the buy-and-build being screened. On record in their own words owning customer acquisition, retention, lifetime value and subscription-mix analysis. And the product is the substitution question in its most current form: paid AI summarisation sold into a market where general AI assistants now do the same thing free.
Check before booking a call
Sale-agreement confidentiality, non-disparagement and any deferred consideration, asked before the call is booked, not after.
Changed during checking
The acquisition date carried by two upstream files was wrong. The company register shows a completion two and a half weeks earlier, and the date the two files used has no supporting source at all; it was struck.
Ranked 3, below the cut line
Retired head of a national schools assistive-technology service
Independent expertAnswers KG1, KG3d, KG2d
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile not found
Confidence and relevance are graded separately.
Confidence: high on tenure and independence; medium on the specific product judgements
Role
Ran the service for decades before retiring in 2025; authored dated comparative guidance for institutions setting free and built-in tools against paid ones
A government core-funded national AT service for schools in the UK, which sells no software and distributes free alternatives
Why they fit
The most deliverable person in the run: retired, no employer, sells nothing. Has been making the free-versus-paid call pupil by pupil for twenty years while the organisation gave the free tools away, and wrote a national exam specification. First substitute for either name above, but explicitly not a witness to the UK student-funding regime, so swapping this candidate in would trade the run’s only first-hand answer on who controls the purchase for a second, better answer on substitution.
Check before booking a call
None disqualifying; disclose a past collaboration with a voice-technology supplier.
Changed during checking
One cited source, a national schools trial of the market leader’s read-aloud product, has since been withdrawn and no longer resolves.
Ranked 4, below the cut line
Divisional commercial leader at the market leader
Former employeeAnswers KG4a, KG1b
Confidence High
Relevance High
Verification Confirmed
Contact Business email found Email found, profile found
Confidence and relevance are graded separately.
Confidence: high on employment; low on the figures
Role
Ran one of the group’s four go-to-market motions globally while colleagues sold to school districts, to 2025
Everway (formerly Texthelp)
Why they fit
The cleanest second witness on whether this is one market or several, and the only one carrying no conflict at all: no board seat, no rival employer, no continuing group relationship. Ran a division that sold in a different motion from the rest of the group, which is exactly the comparison the client needs.
Check before booking a call
Was the remit a P&L, or leading its sales? Two independent sources describe a sales remit and agree with each other.
Changed during checking
No self-reported revenue or retention figure from this file may reach the client. The employer’s own appointment release, recovered from an archive after being withdrawn, describes the remit as leading sales and mentions no P&L, EBITDA or profitability.
Ranked 5, below the cut line
Former senior leader of the market leader’s renewals function
Former employeeAnswers KG1b, KG4a
Confidence Med-high
Relevance High
Verification Confirmed
Contact Professional profile is the only channel Email not found, profile found
Confidence and relevance are graded separately.
Confidence: medium-high: single-source, self-authored, no third-party attestation
Role
Grew into leading the retention function over more than a decade; left at the end of 2024 and is retired with no current employer
Everway (formerly Texthelp)
Why they fit
Knows both the aggregate renewal number and how it was made, which is unusual: most people who can quote the number never worked the accounts that produced it. Explicitly available, no current employer, no rival employer. The account of the renewal book stops at the end of 2024.
Check before booking a call
Did the remit genuinely span workplace and education, and did discount authority sit inside it?
Changed during checking
A 2026 post by the group is attributed to this person by at least one search-engine summary. It was only liked, more than a year after leaving; the misattribution must not travel.
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Has run the institution’s assistive-technology provision since the early 2020s, including the guidance pages that decline the market leader’s licence
A UK research university that publishes it holds no site licence for the market leader, and whose AT catalogue contains no paid literacy suite at all
Why they fit
An institution that publishes, feature by feature, that the market leader is substitutable, and the person who has run the substitute estate that stands in its place. Also raised a threat nobody else in the run identified: academic-integrity rules now barring in assessment the AI-bearing tools the state funds.
Check before booking a call
Did the institution ever hold a site licence and drop it, or never buy one?
Ranked 7, below the cut line
Retired accessibility lead at a Big Four professional-services firm
CustomerAnswers KG2c, KG1a
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on the person; capped at the organisation’s grade, which rests on vendor publication
Role
Set accessibility standards, policy and process for a firm-wide technology estate; retired in early 2025 after nearly thirty years at the firm
A Big Four professional-services firm running a Microsoft-standardised estate
Why they fit
The least constrained candidate in the customer folder: retired eighteen months, explicitly not seeking work. Thirty years of institutional memory of how a large professional-services firm bought and governed assistive technology, in the workplace segment the briefs call the most under-searched and possibly the most defensible part of the category.
Check before booking a call
Does the firm hold a paid licence today, and was anything compared before it did?
Ranked 8, below the cut line
Educational-technology consultant at a Canadian school board
CustomerAnswers KG3a, KG2c, KG1a
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high: customer-published, dated and bylined
Role
In the pedagogical-services team that published the board-wide licence announcement and runs the staff training on the product that replaced the incumbent
A school board in Canada
Why they fit
The best-dated within-category displacement in the run, and it is published by the buyer rather than the vendor: a board-wide licence dated October 2024, force-installed, stating in the board’s own words that it replaces the rival word-prediction product, which survives only as the accredited ministry-exam tool. Two of the three named players in one buying process.
Check before booking a call
In the room when the board decided, or executing a decision taken above?
Changed during checking
Upgraded in verification on the dated board announcement, which the candidate file had not retrieved in full.
Ranked 9, below the cut line
Procurement officer at a large US public school district
CustomerAnswers KG3a, KG2c
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Buyer of record on a competitive re-tender of the district’s incumbent literacy product; signed the notice that names the incumbent
A large US public school district, 30,000+ student users on the incumbent product
Why they fit
A 30,000-seat incumbency of the market leader went to open competitive tender, the district’s own published Q&A names the incumbent product and permits it to bid, and a nine-person challenger bid into it. The award is public but the winner is not retrievable: the board-approval document is a scan with no text layer. This is the person who settles it in one sentence. The risk is availability, not evidence.
Check before booking a call
Can a serving procurement officer discuss an award made a few months ago?
Changed during checking
A specialist reported that the procurement record names no incumbent. Verification pulled the full 41-item Q&A and it does: item 24 names the incumbent product, items 32-33 give the seat count. No contract value may be quoted: the district was asked and expressly declined to give one.
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Principal in a consumer subscription-pricing decision and its public reversal; still in post
A consumer app holding company
Why they fit
A funded acquirer ran a pricing experiment on a disability user base in public and was forced to reverse it in eleven days. Nothing else in the run puts a number on the ceiling of willingness to pay in the consumer half of the category, and this is a principal in the event who will have a defended account of it.
Check before booking a call
Ask about the two price events separately; the earlier one appears to have held.
Changed during checking
Upgraded in verification: there were two price events, not one, and the candidate file described only the second.
Ranked 11, below the cut line
Reading-and-writing-technology researcher at a Danish university college
Independent expertAnswers KG1, KG3d
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on the published record; medium on independence
Role
Long-standing empirical work on what dyslexic students actually do with speech-to-text and AI writing tools, in teacher education
A Danish university college
Why they fit
The substitution researcher the run went looking for and was not sure existed. Published the evidence-based case for specifying by function rather than by programme: the most dangerous idea in this category for an incumbent whose position is a brand on an approved list, and the same idea a Canadian province wrote into policy independently. Not a market analyst, and should not be asked market questions.
Check before booking a call
Project-level vendor funding: two acknowledgements pages, ten minutes, done before delivery.
Confidence High
Relevance High
Verification Reclassified
Contact Below the cut line; contact search not run
Role
Has directed the unit that publishes the institution’s accessibility guidance across a documented non-renewal and every year since
A large US public university that published its own reasoning for not renewing an enterprise literacy licence
Why they fit
The durability witness. The one documented paid-to-free substitution in the category was published by this institution on its own accessibility site, and this is the person who has run the substituted estate through the six years since, the years in which the free substitutes actually got good.
Check before booking a call
Did the decision hold? Nothing in the 2024-26 output mentions a literacy suite.
Changed during checking
Must not be presented as the person who cancelled the market leader. The decision is 2020, outside even the widened window, and the brief’s period disqualifier is flat, so the file was reclassified from cancellation witness to durability witness. Two further corrections: the event was a portfolio-wide licence review rather than a standalone accessibility decision, and the substitute named first was the incumbent platform vendor’s own free educator tier.
Ranked 13, below the cut line
Educational-technology adviser at a Québec school service centre
CustomerAnswers KG3a, KG1a
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Authored a published comparison of two French-Canadian writing-support products, and refreshed it in 2025
A school service centre in Québec that lends student laptops with both products pre-installed
Why they fit
Wrote the best-documented product comparison in the run and then updated it three years later, which closes the “four-year-old evaluation” doubt that would otherwise apply. Sits at an institution that deploys both products side by side, and documents a direct rival going partly freemium.
Check before booking a call
Is the incumbent’s position at this service centre eroding? Interview in French.
Ranked 14, below the cut line
Digital-education policy lead in a Belgian regional education ministry
Independent expertAnswers KG4b, KG2d, KG3d
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Long-standing policy owner for ICT in special-needs education, with a board-level role in a European schools-network body
A regional education ministry in Belgium
Why they fit
The only cross-national vantage point in the run: a view across thirty-plus national digital-education regimes that nothing else here approaches, against a question about whether this is one market or several. Also sits above the regional scheme that funds free reading software against five approved programmes, none of them the market leader’s.
Check before booking a call
In the room for the software approval, or the policy owner above it?
Changed during checking
Two disclosures, neither disqualifying: the job title is literally stakeholder management with the edtech industry, and the European body takes industry partnership funding.
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Ships the free accessibility and AI study features bundled into every M365 Education licence
Microsoft
Why they fit
The free-and-bundled tier represented by somebody who builds it. Do not spend the call on the read-aloud story, which is five years old; the live question is the second wave now shipping: AI summarisation, study agents and content modification free into every education licence, which is the same move that took read-aloud off the table, one product generation on.
Check before booking a call
Employer approvals, and an unwillingness to disparage partners.
Ranked 16, below the cut line
Founder-CEO of a small challenger vendor
CompetitorAnswers KG3c, KG4
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Bid into the competitive re-tender of a 30,000-seat incumbency at a large US school district; four of its scope challenges were carried into the addendum
A small edtech vendor
Why they fit
One of only two documented competitive encounters in the run, and the mirror of the buyer-side candidate on the same procurement: the client can have both sides of one transaction. The district’s Q&A quotes the challenge, and the same-day addendum removed all four contested sections.
Check before booking a call
A serving founder-CEO on a live district outcome; pricing talk will be constrained.
Changed during checking
Only the first question in the procurement record is attributed to this vendor; do not credit it with the remaining scope challenges as established fact.
Ranked 17, below the cut line
Former US regional renewals leader at the market leader
Former employeeAnswers KG1b
Confidence Medium
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: medium: single-source, no third-party attestation of any kind
Role
Player-coach quota, then direct co-management of renewal representatives, then a multi-state renewal book with a team of account managers, to early 2025
Everway (formerly Texthelp)
Why they fit
Filed as the weakest of nine and offered for cutting; the role description settles the doubt in the candidate’s favour and the file stays. No current employer of any kind, which is rare in this folder.
Check before booking a call
Where has this person been since leaving in early 2025?
Changed during checking
Severe namesake risk: match on the employment history, never on the name. No self-reported quota or revenue figure from this file may reach the client.
Ranked 18, below the cut line
Former US renewals manager at the market leader, now at a rival
Former employeeAnswers KG1b
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Nearly seven years on the US renewal book, having previously sat on the buying side of the same conversation
Everway (formerly Texthelp); now at a competing voice-technology vendor
Why they fit
The only renewal-book witness in the run whose account runs into late 2025, which is the year the client cares about; every other vendor-side witness stops in 2024 or early 2025.
Check before booking a call
Did this person carry a renewal book in 2025, or move to presales?
Changed during checking
Candour flag: now employed to displace the product previously renewed, in the same accounts.
Ranked 19, below the cut line
Long-serving assessor at a UK workplace assistive-technology provider
Channel partnerAnswers KG3b, KG2a, KG1
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: strong on the employer; positional on the person Relevance: high or moderate, decided entirely by the screening question
Role
Nearly two decades specifying products for individual workers under the national workplace-adjustment scheme
A UK workplace AT assessment and supply company that carries two rival literacy products built by the same creator
Why they fit
The employer sells the market leader’s acquired reading product and a rival side by side, and the rival was built by the person who created the first one. Two decades of choosing between those two, repeatedly, for real people, is the most direct judgement on that acquired franchise the run reached.
Check before booking a call
Decisive: has this person personally specified one of the two products against the other for the same individual? If not, the file is a long-tenured technician.
Ranked 20, below the cut line
Commercial lead at an independent UK workplace AT reseller
Channel partnerAnswers KG3b, KG2a, KG4
Confidence Med-high
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: medium-high on the current role; medium-low on the earlier two decades
Role
Carries two of the market leader’s literacy products, a dedicated workplace-scheme SKU and a competing dictation product on one shelf; two decades at a student-funding one-stop shop before that
An independent UK workplace reseller
Why they fit
The current seat is the asset: uniquely able to say whether the market leader’s overlapping literacy products compete against each other on the same shelf, which is a buy-and-build question no vendor-side witness will answer straight.
Check before booking a call
Did the former employer hold student-funding one-stop-shop business in the 2002-2020 period, and did this person see its prices?
Changed during checking
Discount the “displaced before 2024” framing until the earlier employer is graded; it rests on the candidate file’s inference, not a source.
Ranked 21, below the cut line
Commercial relationships lead at one of the two UK DSA framework primes
Channel partnerAnswers KG2a, KG3b, KG1
Confidence High
Relevance High
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: strong, downgraded: the current position is now single-sourced
Role
Runs commercial and strategic relationships at the prime, and holds a role at one of its shareholder companies; previously ran a one-stop shop’s student-funding line under the pre-2024 reverse-auction regime
A joint venture holding half of the UK Disabled Students’ Allowance framework
Why they fit
A first-hand account of what the February 2024 transition did to the pre-existing assessment market, from inside the structure that replaced it. Not an independent second reading of the regime, though: a colleague of another candidate in the same joint venture; deliver at most one of the two.
Check before booking a call
Which company actually pays this person; and a live framework interest expiring 19 November 2026.
Changed during checking
Filed as a trade association’s Executive Director. That role had already ended well before the file was written, and a different officer post is held now; the cited source announced a past appointment and was read as current. It was one of the file’s two independent sources, so the current position drops to single-sourced. The description must be struck everywhere.
Ranked 22, below the cut line
Commercial lead at a well-known but small US literacy-software brand
CompetitorAnswers KG3c, KG1
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on employment; positional on the encounter
Role
Runs a brand of roughly seven people inside a private-equity education portfolio
A US literacy-software brand inside a PE-owned education group
Why they fit
Uniquely placed on two things: why the category’s most expensive product sits on a European approved list while the global leader does not, and what a famous brand run by seven people looks like from inside a private-equity portfolio. Do not spend the call on the ownership-boilerplate lead; it deflates, because three sibling brands dropped out of the same document at the same time and one still has a named president.
Check before booking a call
The most constrained speaker in the folder, but reachable: the employer publishes a work address.
Ranked 23, below the cut line
Commercial person inside a UK reseller that co-owns a framework prime
Channel partnerAnswers KG2a
Confidence Medium
Relevance High
Verification Downgraded
Contact Below the cut line; contact search not run
Role
Sat inside a reseller, a framework prime and a catalogue vendor simultaneously across the 2022-2025 transition
A UK assistive-technology reseller and shareholder in one of the two DSA framework primes
Why they fit
Three positions in the UK chain held at once through the whole transition, at the reseller the market map judged best-placed: it carries two of the three named players and co-owns half the framework.
Check before booking a call
What the third company is, and whether the vendor role was employment or a concurrent contract.
Changed during checking
The claim that made this file interesting does not hold. The vendor-side role was demand generation into the specifier network, not management of the catalogue position, so this candidate does not answer what a catalogue position costs a vendor to hold, and was reranked accordingly.
Ranked 24, below the cut line
Commercial contact at the exclusive francophone distributor for one named player
Channel partnerAnswers KG3b, KG1, KG2a
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: strong on the company; strong but unconfirmed on the person Relevance: high, and narrow
Role
In the commercial function at the distributor, carrying the named player’s French-language products alongside the dominant French writing aid
A Québec distributor holding exclusive global rights to one named player’s francophone line
Why they fit
The one commercial contact the run reached inside that channel, able to compare a named player against the dominant French writing aid out of the same bag, a head-to-head that appears nowhere else in the run. Narrow: answers one named player and neither of the other two.
Check before booking a call
Does this person personally recommend to buyers, or support a separate sales function? Interview in French.
Changed during checking
The two names two earlier stages listed as separate Canadian resellers are one company, rebranded in 2024 (confirmed from its own dated history timeline). The candidate still describes the employer by the old name.
Ranked 25, below the cut line
The second principal at the same small acquired vendor
Former employeeAnswers KG4a, KG2b, KG1b
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on facts; medium on the pricing claim
Role
The other person at the company described at rank 2; left two months after completion
A small vendor holding a position on the UK student-funding catalogue, acquired into the multi-brand group in 2024
Why they fit
One vantage point with a choice of person. Do not deliver this file and rank 2 both. Second choice on the evidence: ownership of pricing and the catalogue relationship is inferred from the title here, where at rank 2 it is evidenced in the candidate’s own words. Prefer this file only if the other declines, or if the question is specifically what the acquirer said it intended.
Check before booking a call
Who set the consumer price ladder?
Ranked 26, below the cut line
Commercial director at a Benelux dyslexia-software vendor and reseller
Channel partnerAnswers KG2a, KG3b, KG1
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: strong, and routing is conditional on the screen
Role
Commercial leadership at a company that both builds a product on a national approved list and resells others
A Netherlands-based dyslexia-software company inside a larger group
Why they fit
The run’s only Benelux vantage point, and the only person inside a national scheme the market leader is excluded from. Routing is conditional: if the employer carries no third-party vendor line, this candidate fails the brief’s comparison test and belongs in the competitor population as a rival vendor’s commercial director.
Check before booking a call
Decisive and cheap: does the employer actually carry a third-party vendor line? Ten minutes on its own product pages, resolved before delivery.
Changed during checking
The market map recorded this employer as owned by another named vendor. It is not; the two are sister brands under a common group.
Ranked 27, below the cut line
Non-executive board member of the non-profit that administers a regional approved-software scheme
Independent expertAnswers KG3d, KG4b, KG2d
Confidence High
Relevance High
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on identity; medium on product-level judgement
Role
Involved since the body was founded; now a non-executive rather than its operating head
A very small non-profit in Belgium that is the entire channel for five competing vendors, and sets a file-format requirement they must meet
Why they fit
Excellent on why the scheme exists and on the format dependency it imposes: a format-level barrier to entry set by a charity, and a plausible unpublished reason no product of the market leader’s qualifies. Switching cost in this regime is set by the intermediary, not the vendor: you cannot change product mid-year.
Check before booking a call
This is the counterparty who would have declined a vendor application: a candour constraint, distinct from independence.
Changed during checking
Admitted as vice-chair rather than operating head, which the candidate file had not established. Two disclosures: the body has a standing arrangement with every approved vendor, and an edtech platform was created inside a commercial vehicle. Not evidenced as the person who actually scored the vendor packages.
Ranked 28, below the cut line
Long-serving commercial manager at the voice supplier the market leader names as its counterparty
SupplierAnswers KG1c, KG4c
Confidence High
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on the entity; medium-low on the OEM book
Role
More than two decades in sales at the supplier’s Swedish entity, with OEM, VAR and distribution agreements in the stated remit
A European text-to-speech supplier, owned since 2022 by an adjacent assistive-technology competitor, and a supplier to two of the three named players
Why they fit
Sits at the exact legal entity the market leader names as its counterparty, an entity correction that was load-bearing, because a search scoped to the supplier’s original country finds only engineers. The best available reading of what the incumbent’s cost of voice actually is.
Check before booking a call
Are the supplier’s OEM licences handled by territory sales managers at all?
Changed during checking
The national register entry evidences the legal entity, not a live product; the product it was cited for was discontinued from January 2021. The documented specialism is custom voice branding, with accessibility one of seven markets.
Ranked 29, below the cut line
Education sales director at an embedded text-to-speech supplier
SupplierAnswers KG1c, KG4c
Confidence High
Relevance Medium
Verification Confirmed
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on the role; low on the specific account
Role
In the commercial function that licenses voice into other vendors’ products, covering education, accessibility, e-assessment and LMS buyers across EMEA
A voice supplier that changed private-equity owner in September 2025
Why they fit
The best available witness on how an embedded TTS partner licence is structured and priced, and on whether education, accessibility, e-assessment and LMS buyers are one segment or four, which is the “one market or several” question asked from underneath. The employer has just been through a private-equity diligence on this market.
Check before booking a call
Doubly constrained: supplier to the market leader and competitor to it at the same time.
Changed during checking
Not the account witness the file was built on. The link to the named product rests on one line of one document nobody retrieved, and the file is admitted on the segment limb only.
Ranked 30, below the cut line
Former product leader at the company that merged into the group
Former employeeAnswers KG4a
Confidence Medium
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: medium: the eighteen months that matter rest on a self-authored record alone
Role
A senior product leader at the merging company for a decade, and states an advisory engagement continuing through the merger
The US company that merged into the target group in 2024
Why they fit
If the advisory engagement was substantive, a strong witness to what the merged businesses actually integrated. If it was a consulting tail, a pre-merger product witness, a different and lesser thing.
Check before booking a call
What the eighteen-month advisory engagement actually was, and which side of the merger it advised.
Changed during checking
Not a renewals witness at all, contrary to the filing. Discount for polish: this candidate has pre-packaged the profile as M&A integration advisory.
Ranked 31, below the cut line
Former Nordic product owner at an acquired brand, now at a direct rival
Former employeeAnswers KG4a, KG2b
Confidence High
Relevance Medium
Verification Confirmed
Contact Below the cut line; contact search not run
Role
Two decades as product owner of a Norwegian literacy product acquired into the group; product rather than commercial
A Norwegian literacy-software company acquired into the target group; now at a Nordic rival
Why they fit
The run’s only Nordic witness, in the geography where the market leader’s position is weakest. Value is narrow and real: what four acquired Nordic products became inside one group, and what the national provision route demands of a vendor.
Check before booking a call
Personally ran the product consolidation, or was the product owner while others did?
Changed during checking
Carried no commercial number; do not stretch this file into the renewals gap. Conflict: now at a direct rival competing with the target in all three Nordic languages the interview would cover.
Ranked 32, below the cut line
Former assistive-technology supervisor at a large US school district
CustomerAnswers KG1a, KG3a
Confidence High
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on 2006-2025; medium-low on the link to the tender
Role
Supervised the district’s AT unit for six years, after thirteen years delivering AT there; left the post in 2025
A large US public school district
Why they fit
Nineteen years of how one very large district actually deployed AT, and the coexistence of paid and free tools inside it. Retrospective only.
Check before booking a call
Did the later post retain any district AT portfolio?
Changed during checking
Must not be presented as the district’s AT leader through the 2026 tender; that framing would collapse in ten minutes. The AT-supervising post was left seven months before the RFP issued, for an unrelated early-childhood role. Severe namesake risk: two AT specialists of the same name in the same county.
Ranked 33, below the cut line
Former account manager at an Ontario assistive-devices-programme vendor
Channel partnerAnswers KG2a, KG1
Confidence High
Relevance High
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on employment; low that the product was in the personal book Relevance: high or near-zero: the screen decides
Role
Six years with stated margin responsibility inside a price-controlled provincial regime; has since left the industry, so is free to speak
An Ontario assistive-technology vendor registered under the provincial assistive-devices programme
Why they fit
A price-controlled provincial regime is the cleanest place to see what a channel margin actually is, and this candidate states margin responsibility in their own words. Whether it is worth anything depends entirely on the first question.
Check before booking a call
First question, before anything else: was the named player’s product ever in this person’s personal book?
Changed during checking
The candidate’s own description points the book at communication devices rather than literacy, and both prior employers are vision companies. Self-reported revenue figures from this file must not travel.
Confidence Medium
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: medium on employment; low that the national tender was worked
Role
More than two decades selling into the Danish municipal and school market
A small Danish literacy-software vendor
Why they fit
The municipal half of the Danish picture, which the buyer-side Danish candidate cannot cover. Now inside a live re-tender and correspondingly constrained.
Check before booking a call
Which accounts does this person actually carry?
Changed during checking
Must not be presented as the seller who took the Danish national contract off the market leader; that framing will not survive the first question. A single-supplier national framework at a twenty-five-person company is the chief executive’s work, and the file evidences an individual-contributor municipal seller.
Confidence High
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on identity; low on institutional exposure
Role
Worked on the browser’s built-in reading and image-to-text features; left in 2025
Google
Why they fit
Can say why a browser vendor built read-aloud free and how that was justified internally: the supply side of the substitution question, from someone who shipped it.
Check before booking a call
None, but this candidate cannot answer what institutions did.
Changed during checking
No source shows any contact with a school district or institutional buyer. The stated record is consumer AI at scale, so the file is admitted on the supply side only.
Confidence Medium
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Role
Nine months in post at the one institution running two of the three named players side by side
A large US public university holding a site licence for one named player and a capped pilot of another
Why they fit
Real operator-level colour on student sentiment where two of the three named players are deployed at once. Not the renewal conversation.
Check before booking a call
Does the student-services function have any role in allocating pilot seats?
Changed during checking
Not in the IT accessibility group, the team that actually holds both the site licence and the pilot. Filed as a witness to a procurement seat this candidate does not occupy.
Confidence Medium
Relevance High
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: medium on employment; low that the literacy product was carried Relevance: high or near-zero: the screen decides
Role
Sales role at a vendor whose primary identity is visual impairment; the role entry carries no products, territory or segment
A Belgian AT vendor with a product on the regional approved list
Why they fit
The finding about the Flemish approved list is confirmed and sharp: the market leader is on none of the five approved programmes, and the reason is procurement rather than language. This is a weak carrier for it.
Check before booking a call
Open by asking what this person sold. If the answer is low-vision hardware, the brief excludes the file and the call should end.
Changed during checking
The clinical training is outside literacy and the employer’s primary identity is visual impairment, which the brief excludes outright. A better-aimed person at the same employer was found and named as a replacement rather than a supplement.
Confidence High
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on employment; low on account ownership
Role
Long tenure at the supplier, latterly in its inclusive and assistive segment
A European text-to-speech supplier owned by an adjacent assistive-technology competitor
Why they fit
Admitted on the segment limb only: whether accessibility buyers behave as one market or several from the supplier’s side.
Check before booking a call
None that rescues the original pitch.
Changed during checking
The reason this file existed does not survive. It was built as the person who watched two of the named players buy the same input; the candidate’s own headline and achievements name voice preservation, AAC, healthcare and charity partnerships, not OEM licensing to software vendors. Ranked below the other candidate at the same supplier, reversing the specialist’s own order.
Ranked 39, below the cut line
Digital-skills officer at a UK fire and rescue service
CustomerAnswers KG2c, KG1a
Confidence Low
Relevance Medium
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: low on current status; no dated employment record found at all
Role
Named against a 2024 enterprise-licence event at the service; current status undocumented
A UK fire and rescue service holding an enterprise licence and a family tier
Why they fit
A rare public-sector workplace account where the licence is confirmed by the buyer rather than the vendor. The account went up in verification and the person went down.
Check before booking a call
Is this person still at the organisation?
Changed during checking
Another named person now holds the title. Two employees of the same organisation have since publicly confirmed the licence and the family tier on their own accounts, so the account no longer rests on vendor publication, but both are better evidenced than this candidate and were recorded as replacements.
Confidence High
Relevance High
Verification Downgraded
Contact Below the cut line; contact search not run
Confidence and relevance are graded separately.
Confidence: high on 2005-2024; unresolved 2024-2026 Relevance: high, and unreachable
Role
Sold synthetic voice into UK education for the best part of two decades, to 2024
A UK text-to-speech supplier named under several of the market leader’s products
Why they fit
Ranked last on reachability, not on merit. The only supplier-side person whose vendor relationship is corroborated by the buyer’s own technical documentation as well as by a statutory register, and unfindable, because three real, checkable and entirely unrelated people share the name. The route in is the company register, never a name search.
Check before booking a call
Unfindable by name. Route in through the statutory officer record or the acquiring company.
Changed during checking
The supplier’s UK entity was struck off in 2026, but it was a voluntary strike-off filed by the company itself after the product migrated to a US parent, which is the benign reading, and it is not evidence that this person has left.
New project
Start a new search
Searches are run from Claude Code, not from this site. A project appears here once its run
has finished and its results have been published.
Steps
- Create a run folder,
projects/<date>-<name>/, and save the client brief in it as 00-brief.md. - In Claude Code, run the
orchestrator skill on that folder. It works out which stages are owed and runs them, pausing once after the expert types are chosen so you can review them. - The last stage writes this site's data file for the project. It then shows up as a new project above.