The research that told a roadmap 'not that' before six months were spent
Jobs-to-be-done interviews and demand signals redirected a planned vendor-risk module toward the problem customers actually had, expense reconciliation, before a single sprint was spent building the wrong thing.
What the engagement involved
- Industry
- Professional Services
- Duration
- 5 weeks
- Cooperation model
- Fixed price
Client name withheld under NDA. Engagement details are shown to the extent our agreement permits.
The question we were asked
The roadmap had a vendor-risk module penciled in as the next big build, based on two enterprise prospects who'd asked for it in sales calls.
A B2B product company with two quarters of engineering time about to be committed to a vendor-risk module, on the strength of two enterprise prospects who had asked for it in sales calls. Leadership wanted evidence before committing, which is an unusually good instinct and an unusually cheap one to satisfy: five weeks of research against six months of building.
Product & Market Research
The decision, first
- 01
The planned feature was a loud ask from a narrow segment; the real one was never requested directly.
Two prospects asking loudly is a real signal about two prospects; the market signal was in what ten of twelve raised without being asked.
- 02
Coding for the job, not the feature, is what surfaced a pain point nobody named.
Buyers name features and describe jobs, and the two rarely match. Coding for the job is what let a pain point nobody requested become the recommendation.
- 03
Five weeks of research replaced six months of building the wrong module.
The saving is asymmetric: research that confirms the plan costs five weeks, and research that redirects it saves two quarters, so the expected value isn't close.
What the numbers couldn't answer
Before committing two quarters of engineering time, leadership wanted evidence that the broader market actually had this problem, beyond the two loudest accounts in the pipeline.
The evidence behind the roadmap decision was two conversations, both with prospects at the top of the pipeline, both mediated by a salesperson with an interest in the answer. That's a real signal about two accounts and no signal at all about a market. Nobody had asked whether the ask generalized, and the roadmap had no mechanism that would have surfaced it if it didn't.
We ran jobs-to-be-done interviews across a dozen prospective and current buyers, pulled demand signals from search and adjacent tool usage, and mapped the market gap. Vendor risk turned out to be a niche ask from a narrow segment, while expense reconciliation showed up as an unprompted pain point in nearly every conversation and had a clear, underserved gap in the market.
How we worked it through
Ran jobs-to-be-done interviews with 12 prospective and current buyers
Interviews were anchored on the last real occurrence of the problem, because recalled behavior is evidence and predicted behavior isn't.
Pulled demand signals from search volume and adjacent-tool category trends
Search volume and adjacent-tool trends were pulled independently of the interviews, so the two signals could confirm each other or fail to.
Mapped the market gap across existing competitors' feature coverage
Competitor coverage was mapped feature by feature, which separates an unmet need from one already answered four times over.
Delivered a research brief recommending expense reconciliation over vendor risk
The brief named a specific gap (the multi-currency case buyers kept describing) instead of a category, which is what made it buildable.
Phase by phase
Phase 1: Interview
Twelve buyers, one question
Ran jobs-to-be-done interviews with twelve prospective and current buyers, coding for the underlying job.
- Interview guide
- Coded transcripts
Phase 2: Size
Signals outside the room
Pulled demand signals from search volume and adjacent-tool category trends to size each candidate independently.
- Demand analysis
- Category trends
Phase 3: Map
Where the gap actually is
Mapped competitors' feature coverage to separate an underserved need from a well-served one.
- Competitive map
- Gap analysis
Phase 4: Recommend
Say the unwelcome thing
Delivered a research brief recommending expense reconciliation over the planned vendor-risk module.
- Research brief
- Recommendation memo
Four recurring jobs: every card a sentence a buyer said, sorted into the job it belongs to, with vendor risk alone in the last column and the two mentions that came only after a roadmap question shown apart.
What it changed
10 of 12, unprompted
Interviews that surfaced the real pain point
~2 quarters
Engineering time redirected before being spent
Expense reconciliation, underserved
Market gap identified
Ten of twelve is a count of interviews in which expense reconciliation surfaced without being prompted. The redirected engineering time is the module's own estimate, not a measured saving: it's time not spent, which isn't the same as time recovered. The market gap is a judgment supported by the competitive map, and is reported as one.
Client name withheld under NDA. Figures are approximate, drawn from the engagement’s own reporting.
Interviews, not a survey
Twelve buyer conversations coded for the job being done, whatever feature was requested.
Twelve conversations, each coded for the job the buyer was hiring a tool to do, set apart from the feature they asked for. The two are routinely different, and the feature request is the less useful of them. Interviews were structured around the last time the problem occurred, never around hypotheticals, because recalled behavior is evidence and predicted behavior isn't.
- Twelve interviews coded for the job, not the feature request
- Anchored on the last real occurrence, not on hypotheticals
- Same coding frame applied to prospects and current buyers
Who we spoke to: all twelve buyers, six current and six prospective, with sector, relationship, the last real occurrence each interview was anchored on and how many jobs the transcript coded to, beside one transcript coded for the job, with the feature it named kept but not counted.
Demand outside the room: two years of search interest and adjacent-tool category trends pulled independently of the interviews and drawn to shape, then checked against them. Expense reconciliation was large and growing in both; vendor risk was the smallest term on the page.
Demand signals
Search volume and adjacent-tool trends used to size the ask independently of what buyers said.
What buyers say and what they search for are separate signals, and agreeing is what makes either trustworthy. Search volume for the two problem areas and usage trends in adjacent tools were pulled independently of the interviews, then compared. Expense reconciliation was large and growing in both; vendor risk was loud in three conversations and near-invisible everywhere else.
- Demand sized independently of what interviewees said
- Search volume and adjacent-tool usage compared against interview coding
- Agreement between the two treated as the test
A mapped gap
Wanted often, served badly: six candidate jobs plotted by how often buyers raised them against how many tools already do them fully, vendor risk in the corner four tools own, and the feature-by-feature coverage map that names the gap: multi-currency card matching.
Competitor feature coverage mapped, so an underserved need could be distinguished from a crowded one.
Competitor coverage was mapped feature by feature so an unmet need could be told apart from a need already met four times over. Vendor risk was crowded; expense reconciliation had partial answers with a specific gap: nobody handled the multi-currency case the buyers kept describing. The recommendation named that gap instead of the category, which is what made it buildable.
- Competitor coverage mapped feature by feature
- Crowded needs separated from genuinely underserved ones
- Recommendation named a specific gap, not a category
How the engagement ran
- 01
A cross-functional team of 3 worked on a fixed price basis over 5 weeks, covering JTBD interviews, Demand signal analysis, Market gap analysis. We ran a weekly demo and a shared board they could read at any time. Their team took over day-to-day operation before the engagement ended, with handover built into the last phase.
Five weeks, fixed price, ending in a brief instead of a build. Interviews were anchored on the last time the problem actually occurred, not on hypotheticals, and were coded for the job the buyer was hiring a tool to do, not the feature they named. The demand analysis ran independently of the interviews so the two could agree or fail to.
Twelve interviews, four recurring jobs
The pain nobody requested was the one nearly everybody described
Pick a recurring job to see which of the twelve buyers raised it unprompted, which only when asked, and which never did. Below it, the three tests that put expense reconciliation ahead of vendor risk. Switch tabs, or use the arrow keys once one is focused.
Twelve interviews, one recurring job at a time
- INT-01ControllerCurrent · Engineering consultancy
- INT-03Head of financeCurrent · Staffing firm
- INT-05Accounting leadCurrent · Architecture practice
- INT-08ControllerCurrent · Law firm
- INT-10Senior accountantCurrent · Accounting firm
- INT-12ControllerCurrent · Construction
- INT-02Finance managerProspective · Marketing agency
- INT-04VP financeProspective · Healthcare group
- INT-06CFOProspective · Software company
- INT-07Procurement directorProspective · Manufacturing
- INT-09Finance directorProspective · Nonprofit
- INT-11Head of procurementProspective · Logistics
“Every close, someone rebuilds the exchange rate for each card line in a spreadsheet.”INT-01 · unpromptedOnly reconciliation’s 10 of 12 and vendor risk’s three conversations are the study’s figures; the rest of the coding is illustrative.
Why expense reconciliation won over vendor risk
From two loud sales calls to a gap the market actually has
Three independent lines of evidence (what buyers described, what the market searches for, and what competitors already cover), each able to disagree with the others, ending in a brief instead of a build.
- 01 · SourceTwelve buyer interviewsProspective and current buyers, each anchored on the last real occurrence of the problem, because recalled behavior is evidence and predicted behavior is not.
- 02 · CodeJob, not featureEvery transcript coded for the job the buyer was hiring a tool to do, with the same frame applied to prospects and current buyers.
- 03 · SizeSearch + adjacent-tool trendsPulled independently of the interviews, so the two signals could confirm each other or fail to. Agreement is the test.
- 04 · MapCompetitor coverageMapped feature by feature, which separates an unmet need from one already answered four times over.
- 05 · DeliverResearch briefNames a specific gap (the multi-currency case buyers kept describing) instead of a category, so the recommendation is buildable.
So the answer isn’t the one you walked in with
Evidence that could have said "build it," and didn't
Nobody led the witness
Interviews were anchored on the last time the problem actually occurred and coded for the job, not the feature. A pain point counted when the buyer raised it unprompted, and ten of twelve did for reconciliation.
Signals able to disagree
Demand was pulled from search volume and adjacent-tool trends independently of the interviews, then compared. Vendor risk was loud in three conversations and near-invisible everywhere else, and that disagreement was reported.
Figures labeled for what they are
Ten of twelve is a count. The ~2 quarters is the module's own estimate: time not spent, not a measured saving. The market gap is a judgment supported by the competitive map, and is reported as one.
Is your next build backed by the market, or by the two loudest accounts? Scope your build in 3 minutes.
Scope your buildNearby engagements
AI & AutomationA private legal assistant grounded in verified precedents
A private knowledge assistant that searches internal case files and precedents, providing cited answers legal teams can verify in seconds.
Legal & Law Firms · 14 weeks
Product DesignAn onboarding flow that guides trial users to value
A redesigned SaaS trial onboarding experience with progressive checklists, sample data, and inline guidance that turns signups into active subscribers.
Professional Services · 10 weeks
Product DesignA design system that brought speed and consistency to 4 product teams
A token-based design system in Figma and React that eliminated component duplication across 4 product squads and cut the time from design handoff to merged frontend.
Professional Services · 14 weeks
Let's talk
Running a large platform, shaping a first MVP, or getting a product ready for a funding round? Tell us where you are. We'll shape the process around it, and stay with you after launch.
