Strategy
Part of Small Business Market Research: Trends, Cohorts, Competitors
How to Check a Small Business Market Research Example Before You Act on It
How to check a small business market research example before acting on it: the source, unit of count, geography, reference period, sample, question wording and the pass mark to set.
What to take away
- Read six things before the conclusionthe decision, the source, the unit of count, the geography, the reference period and the threshold.
- Official counts set a denominator, not demand. US Census County Business Patterns, the BLS Quarterly Census of Employment and Wages and Statistics Canada's Canadian Business Counts all measure locations and jobs.
- Behavior evidence outranks stated intent. Paid pilots, bookings and lost-quote interviews resist flattery.
- Every useful example ends with a pass mark. If no result would change your decision, the example is decoration.
A small business example often looks convincing because it quotes a figure, but the figure alone tells you little. This article sets out the checks to run on any example, the sources to trace each number to, and how to set the pass mark before you act. The two worked examples are hypothetical and labelled as such.
How to read an example in five checks
- Name the decision. A lease choice and a hiring choice need different evidence.
- Trace the number to its source, unit and reference period. Employer locations are not households.
- Match the geography to your service area. A county total says little about a 12-minute drive time.
- Split context evidence from behavior evidence. Population counts are context; invoices are behavior.
- Demand a threshold. Write the number that would change your mind before you start.
Many published examples stop at step four. They describe a method, quote a figure, and never say what result would justify action. Test the choice set the same way, comparing each real alternative a customer could pick.
Survey checks an example should pass
A survey-based example should also report its method. Check the sample design and the base size behind each percentage. Check the margin of error, if one is given, and whether it applies to the subgroup you care about.
Check the exact question wording, the order of questions, whether weighting was applied, and who sponsored the work. A sponsor who sells the result has an interest in it. If an example gives a percentage with no base, treat it as unverified until you have the base.
Hypothetical example: sizing a US county before committing to a second site
A tutoring operator with one location weighs two school districts. Census County Business Patterns gives establishment counts by county, industry code and employment size class. The BLS Quarterly Census of Employment and Wages adds quarterly employment and wage levels. Both sources are public, dated and free.
Counts settle supply, not demand. They show how many tutoring establishments already operate in each district and how those firms cluster by size. Wage data show what a qualified tutor costs in the same county, which sets the floor for any pricing test.
Six factors then decide the district: school-age population, household characteristics, travel routes, competing providers, school calendars and where current inquiries come from.
Tutoring Location Site Criteria
- School-age population
- Household characteristics
- Travel routes
- Competing establishments
- School calendars
- Existing inquiry origins
The pass mark for this example is a count of paid enrollments at the posted price, from a fixed number of open evenings, inside a fixed period. Agree the count and the deadline before the evenings start. A weaker result sends the second site back to review.
Hypothetical example: reading employer size bands before adding a crew
Statistics Canada's Canadian Business Counts table reports employer business locations by geography, employment size band and industry for a stated reference period. Its notes define the unit of count. Check those notes before you compare two editions.
The size band you sell into shapes whether a second crew is viable. The count shows how many firms sit in each band, but not what they buy, so test buying behavior in that band with quotes before committing.
A second crew raises fixed cost before it raises revenue, so the firm needs its own quote log to decide. Track lost quotes, travel time per job and margin per visit, and set the pass mark before the crew starts.
Check for a retired or replaced series
Agencies retire and replace series, and the replacement often uses a different survey, scope or classification. Before you use a figure, record its reference period, industry and geography, and whether a replacement exists, with that replacement's coverage. When the replacement draws on a different survey, the two numbers are not comparable, so do not splice them together.
Behavior evidence does the rest: foot traffic counts at fixed hours, intercept interviews with permission, competitor price mapping and a bounded pop-up. Price the pilot at list and read the conversion, rather than buying volume with a deep discount that cuts margin.
Worked outputs: evidence into a pass mark
| Example | Context evidence | Behavior evidence | Pass mark set before the test |
|---|---|---|---|
| Tutoring site | Establishments by size class, district enrolment | Paid enrolments from open evenings | A count and a deadline agreed in advance |
| Second crew | Employer count bands in the service area | Jobs won at list price | Won jobs per month at target margin |
| Retail pop-up | Regional turnover, footfall by hour | Pop-up takings per trading hour | Takings that cover rent at the observed footfall |
| Recurring offer | Invoice history, renewal dates | Signed pilots of a fixed length | Share of pilots that move to longer terms |
The rows are illustrative decision rules for these scenarios. They are not published averages, and no threshold here is a benchmark. Set your own before the test starts.
Two questions finish the read. Who counted the unit, and does the period still run?
Common questions
Do official counts prove a market exists?
No. They establish a count for a stated unit, geography, classification and period. They say nothing about reachable buyers, current need, price or competitor response.
How many interviews make an example trustworthy?
That depends on the decision. For narrow business-to-business questions, 10 to 20 conversations per segment is a working range rather than a rule, and the pattern matters more than the total.
How do I spot an invented example quickly?
Look for a missing unit, date or threshold. Clear writing about method with no source, no population and no pass mark is illustration, not evidence.
Can I reuse another firm's numbers?
Only when the unit, geography, period and definition match yours. A state retail figure and a single-street pop-up are different measurements.
