Start with decisions, not a market-size slide
| Decision | Research question | Evidence needed |
|---|---|---|
| Choose the buyer | Who feels the cost, champions change, and controls budget? | Recent workflows, org chart, procurement path, existing spend |
| Choose the job | Which recurring event creates enough urgency to change behavior? | Incident examples, frequency, consequence, workaround |
| Choose the wedge | Which segment, geography, workflow, or buying model is underserved? | Competitor focus, reviews, pricing, product gaps, regulation |
| Choose the channel | Where can this buyer be identified and reached repeatedly? | Lists, communities, partners, search terms, media targeting, acquisition cost |
| Choose the price | What value, budget category, and alternative anchor shape the trade-off? | Current spend, labor cost, loss exposure, approval threshold |
| Choose the next test | Which remaining uncertainty could still stop the build? | Explicit assumption, rejection evidence, cheapest falsification method |
Build a source stack with different failure modes
- Recent-behavior interviews: rich workflow and language, but small and vulnerable to courtesy bias.
- Product reviews and support discussions: repeated failure language, but users and context may be unqualified.
- Competitor pages, demos, pricing, terms, and changelogs: category and offer truth, but not competitor traction.
- Public filings, tenders, regulation, and industry bodies: budget, risk, and mandatory workflow context, but often broad or delayed.
- Search and community language: observable attention and vocabulary, but no proof that your proposition or price converts.
- Lists, partners, panels, media planners, and owned audiences: reach and cost clues, but quality must be tested in the intended cohort.
Run the research in six passes
Write a segment hypothesis
Name role, company context, geography, recurring event, economic consequence, current alternative, and plausible channel.
Map the buying system
Separate user, champion, economic buyer, procurement, security, legal, implementation, and renewal roles where they differ.
Study alternatives
Include direct software, adjacent tools, spreadsheets, staff, agencies, avoidance, and doing nothing. Record public price and switching friction.
Collect recent incidents
Interview or survey qualified people about the last occurrence, sequence, consequence, workaround, spend, and failed attempts.
Test reach before scale
Estimate whether the cohort can be identified in an owned list, partner channel, panel, community, search result, or paid audience without broad leakage.
Write the rejection test
State which missing pain, buyer, source, price, trust, or action evidence would stop investment before choosing the next method.
Keep an evidence ledger instead of a confidence score
| Field | Record | Why it matters |
|---|---|---|
| Claim | One falsifiable sentence | Stops a collection of links becoming a conclusion |
| Evidence | Observation, quote, price, behavior, or document | Preserves what was actually found |
| Source and date | Exact URL, participant class, or record date | Makes recency and provenance inspectable |
| Evidence type | First-party, primary external, vendor claim, proxy, or inference | Prevents unlike signals from being averaged |
| Counterevidence | What points the other way | Reduces confirmation bias |
| Confidence | Low, medium, or high with reason | Makes sample and source quality visible |
| Decision impact | Continue, refine, reject, or test | Connects research to the next cost |
Avoid a precise internal score unless every input has a defined scale, provenance, and decision role. The pain, purchasing power, reach, and founder-advantage screen is a ranking aid; it does not turn research into validated demand.
Know when desk research has done enough
- You can name the buyer, painful event, current alternative, budget path, and a reachable source without generic labels.
- You know which competitor or workaround sets the trust and price anchor.
- The founder can deliver or build a focused outcome if demand earns the next gate.
- One uncertainty remains capable of stopping the build, and a priced proposition can test it.
- The action is meaningful, truth-labeled, operationally fulfillable, and measurable.
At that point, more desk research can become avoidance. Use the pre-build validation protocol to define qualified exposure, price, action, budget, and decision rules before traffic arrives.