Costs
Part of Startup Market Research: A Practical Guide
Common startup market research questions, ranked
Common startup market research questions, ranked from the recurring problem to the customer, alternatives, price, and channel.
Common startup market research questions, in order
- Does the problem exist and repeat?Confirm there is a recurring need before investing in a solution.
- Who exactly has it?Define the customer segment after establishing the need, rather than guessing who might buy.
- How do they solve it today?Map the alternatives and workarounds for that segment to see what a new offer must improve.
- What will they pay?Assess willingness to pay once the customer, problem, and current alternatives are clear.
- Which channel reaches them?Test how to reach the defined customer after the offer is specific, before committing spend.
- Write the disclosure list before fieldwork, not aftersponsor, population, sample, dates, size, weighting, and limitations.
- Sequence research against commitments. Interviews before the offer is specific produce opinions about nothing.
- Interviews stop when new segments stop changing the next action, not at a round number.
- Size the market as low, base, and high cases and name the assumption that moves the answer most.
- Stop a cycle when the next step is reversible and another study would not change it.
Disclose the study before you run it
The AAPOR survey disclosure standards list what a reader needs to judge a result:
Survey disclosure checklist
- Sponsor named
- Instrument described
- Population defined
- Sample and recruitment
- Mode and field dates
- Size and weighting
- Processing and quality
- Limitations stated
Startups skip this because nobody asked. Then a board deck cites "78% of respondents" with no denominator, and the number cannot be defended in the meeting where it matters.
Write the disclosure list first. It forces the sampling question early, while the design is still cheap to change.
Time each study against the commitment it gates
Research earns its cost by sitting in front of a decision that is expensive to reverse. Place it there.
Match study to commitment
Commitment
- Build or inventory
- Problem repeats
- Hiring
- Work volume, shape
- Paid media
- Message, audience
- Pricing
- Willingness to pay
Settle first
- Build or inventory
- Interviews, observation
- Hiring
- Time-and-motion, tickets
- Paid media
- Landing page test
- Pricing
- Van Westendorp, offer tests
Study that fits
- Build or inventory
- Hiring
- Paid media
- Pricing
What to settle first
- Build or inventory
- Problem exists and repeats
- Hiring
- Work volume and shape
- Paid media
- Message and audience
- Pricing
- Willingness to pay
Study that fits
- Build or inventory
- Interviews, observation
- Hiring
- Time-and-motion, existing tickets
- Paid media
- Landing page test, small spend
- Pricing
- Van Westendorp, offer tests
Secondary research sets vocabulary and market boundaries. It is cheap and it runs first. Interviews and observation clarify the problem. Behavior tests wait until the offer is specific enough that a stranger can react to it.
A test run before the offer is legible measures the offer's clarity, not the market.
Stop interviews when segments stop moving the decision
There is no universal interview count, and anyone quoting one is quoting their own study. Recruit across the segments that could change the decision, then keep going until new patterns stop changing the next action.
Report who was in the sample, who was missing, how you recruited, whether you paid, and what the sample cannot support. A founder's ten friends are a convenience sample; say so.
When deciding which variables to measure, use startup marketing metrics to choose the ones that can move a decision.
Size the market in three cases
Build the estimate from units you can defend:
- eligible customers
- problem incidence
- purchase frequency
- price
- serviceable share
- reachable share
- delivery capacity
Run low, base, and high cases. Then find the assumption that swings the answer most, usually incidence or reachable share, and go measure that one.
A single-point market size hides its own error bars. Three cases show the reader where the estimate breaks.
Treat competitors as evidence of alternatives, not of demand
Competitors prove alternatives exist. They do not prove the startup can win, and they do not prove every competitor is healthy.
Mapping alternatives is real market research, done before any interview. Map direct products, internal workarounds, outsourcing, delay, and doing nothing. Separate documented facts from marketing claims, and mark private performance as unknown rather than assuming it.
Keep search interest inside its stated scope
Google Trends returns sampled, normalized, aggregated data scaled from zero to one hundred. It is not a scientific poll and not an exact count of searches, people, or customers.
Use it for language and relative interest. Corroborate with official counts, recent customer behavior, channel tests, and transactions before it enters a forecast.
Close the cycle with a decision record
The GAO evaluation design guide ties evaluation questions to evidence needs and design choices. Borrow that discipline: name the question, the evidence that would answer it, and the design that produces that evidence.
The W3C Privacy Principles statement gives shared privacy concepts for web systems and warns against shifting privacy work onto individuals. Apply it to your study design, then confirm the governing law and configuration with counsel. Privacy and advertising rules vary by state and country; the FTC, state attorneys general, and the Quebec CAI each set their own.
Stop a cycle when the evidence supports a reversible next step, the key uncertainty is written down, and another study would not change the immediate decision. Kill the idea when a predefined critical assumption fails. Set a trigger to revisit when price, regulation, market conditions, or product scope moves.
Common questions
Should a startup research before building?
Yes, before commitments that are expensive to reverse, and after writing the specific assumption that evidence could change. Research pointed at no decision produces a report nobody reads.
Are competitors proof of demand?
They show that alternatives exist, not that every competitor is healthy or that the startup can win. Study the workarounds and the do-nothing option alongside the direct products.
Should search interest stand in for market demand?
No. Treat it as a bounded signal about language and relative interest, then corroborate it with official counts, recent customer behavior, channel evidence, and transactions.
