Costs
Part of Startup Market Research: A Practical Guide
Privacy requirements belong at the start of a market research checklist, not the end
Startup market research checklist for 2027 covers decision, boundaries, sources, recruitment, surveys, signals, scenarios, behavior, synthesis, privacy, and release.
What to take away
- Set privacy requirements before you choose a research tool. Notice text, consent capture, retention period and deletion trigger come first.
- Run the checklist in orderdecision, secondary data, alternatives, interviews, survey, market size, synthesis, release.
- Every material conclusion needs a traceable source or a coded research record.
- Release only when the decision owner accepts the stated limits and the next review date.
- Keep research data out of marketing lists. That separation is the step most often skipped.
The checklist in sequence
Work top to bottom. Each step names an output and an owner.
- Write the decision the research must support, plus the alternatives, the owner, the deadline and the cost of being wrong.
- Set privacy requirements before you open a form tool, a recorder or a spreadsheet.
- Define scopebuyer, user, payer, geography, industry, use case and explicit exclusions.
- List the founding assumptions and the evidence that would weaken each one.
- Gather secondary evidence from official statistical and regulatory sources. Save dataset, field, units, geography, measured period and release date.
- Map alternativesdirect products, workarounds, outsourcing, delay and doing nothing.
- Interview recent buyers, non-adopters and negative cases. Document screening, incentives, consent, recording and missing groups.
- Pilot the survey before fielding it. Report instrument, recruitment, mode, dates, sample, weighting and limits.
- Build low, base and high market cases, each assumption carrying a cited source.
- Synthesize, tag every finding by source and confidence, then hold the release gate.
Mapping alternatives is real research, done before any interview. The alternatives-first approach explains how that step reshapes the interview script. For the competitor side, the SBA's market research and competitive analysis guide covers how small teams do this cheaply.
Privacy steps that come before tool selection
NIST's guide to using the Privacy Framework 1.1 lists current and target profiles, gap analysis, priorities and progress measures. Adapt that sequence to research purpose, collection, access, retention, vendors and deletion.
Privacy framework adaptation steps
- Research purpose
- Collection
- Access
- Retention
- Vendors
- Deletion
Write these down before any tool is bought:
Privacy steps before tool selection
- Notice textwho collects, why, what is stored, who else sees it.
- Consent capturean affirmative action, its timestamp, and the version of the notice shown.
- Access and correction route for participants.
- Retention period per data type, with a named reviewer.
- Deletion triggerdate reached, purpose met, or participant request.
The W3C Privacy Principles statement warns against shifting privacy work onto individuals. If a participant has to hunt for a deletion route, the checklist failed at step two.
Capture consent before recording, never after. Store recordings apart from marketing contacts. Write an actual retention number and tie deletion to a trigger.
Twelve to 24 months is a common range for interview recordings. Whatever you choose has to be defensible in writing.
Californian contacts carry extra rules on reuse. The CCPA and CPRA email and ad targeting rules cover what you may keep for promotion after research ends.
Example: one market-size calculation and one coded record
Say the decision is whether to sell scheduling software to independent coffee shops in one US metro. Eligible units come from the state license file: 4,200 cafes with on-premise seating. That is a typical figure for a mid-size metro, and you must replace it with your own count.
Incidence comes from your screener, not a vendor deck. Suppose 35% of those cafes still book staff shifts by phone. Frequency is 12 shift changes a year, and price is $45 per month.
Annual value is 4,200 x 0.35 x 12 x $45, about $793,800. Move each input one at a time to build the low and high cases.
Code the record so a reviewer can follow it: R-014, source 'state license file, 2024 release', method 'record count', limits 'includes seasonal stalls', confidence 'high'.
The arithmetic has to be reproducible by a second person. The public-data file method is what makes the number defensible when someone challenges it.
Gates that block release
| Gate | Required record | Block release when |
|---|---|---|
| Decision | Owner, alternatives, threshold | No action can change |
| Method | Population, source, instrument | Inference exceeds design |
| Protection | Notice text, consent record, retention date | Use is unexpected |
| Release | Tagged sources, stated limits, review date | A material claim is not traceable |
Run the gate before the memo reaches the decision owner. A second reviewer reproduces the market-size calculation and challenges the strongest claim.
Keep research distinct from promotion, as the ESOMAR and GRBN online research guideline requires. Participant lists belong in the research file, not the CRM.
