The Department for Business, Innovation, Science and Trade tracked 1,300 UK businesses for four years. Only 53 of the 427 that started out growing kept growing every year. That is the backdrop for every new business plan, and it is why the first question is not what to build. It is whether anyone will pay.
This guide sets out the First Customer Test for UK founders: ten structured interviews, one offer, one ask for commitment, and a decision. It draws on the Small Business Survey 2025 panel report, published on 15 October 2026, and on Rob Fitzpatrick's interview method from The Mom Test, which we credit and build on with a UK legal and pricing layer that generic guides skip.
A note on what this test can do. No survey proves that validating first guarantees success. What the data shows is narrower: growth is uncommon, forecasts usually miss, and the habits linked to better outcomes are testable cheaply. That makes a two-week test a sensible first investment. It does not make it a magic formula, and we will not pretend otherwise.
What four years of UK panel data actually shows
The panel report follows the same 1,300 businesses through the 2022 to 2025 surveys. Its value is that it connects what firms said they would do with what happened next. Three findings matter for a new founder.
Sustained growth is the exception. The report tracked 427 firms with ten or more employees that had grown their workforce in the year before the first survey. Of those, 53 grew in every year to 2025. That is 12.4% of an already successful group, not of all UK firms. Growth also came and went: 47.1% of firms that cut staff in 2024 hired again in 2025. One good or bad year predicts little.
Forecasts ran ahead of results. Depending on the year, between 42.8% and 52.0% of firms that expected to add staff did so the following year. Turnover expectations did better, between 52.9% and 56.7%. Even owners who know their businesses well are wrong about the next twelve months roughly half the time. A founder with no trading history should assume their own forecast is at least as fragile.
The habits linked to better results are cheap to test. Firms that trained staff in 2022 were more likely to have added staff by 2025 (27.5% against 21.3%). Firms that had innovated were more likely to report higher turnover (43.5% against 39.4%). The report calls these associations, not causes, and the panel is not representative of all UK businesses. But each habit has a low-cost version a founder can try: one workshop, one new offer, one mentoring session.
The level of ambition is not the problem. The GEM 2024/25 UK National Report found 36% of working-age adults running a new business or planning to start one within three years, the highest since the survey began in 1999. A Simply Business survey of 1,842 UK small business owners in summer 2026 found 41% citing weak or uncertain consumer demand as a barrier. Plenty of people start. Demand is where they struggle. The test below attacks that directly.
Check the rules before you talk to anyone
Two UK-specific checks come first, because getting them wrong can cost more than a failed idea.
If you hold a Student visa, you cannot trade. The GOV.UK Student visa guidance says you cannot be self-employed, and UKCISA's guidance confirms you cannot run a business at all with Student permission. Interviews, research and planning are generally treated as pre-business activity. The line moves once money or supply is involved. A signed letter of intent does not change your visa position: it records interest, and if that customer then pays you or you start delivering, you may be trading. Our guide to starting a business in the UK as a student covers the routes and timing. When in doubt, ask your university's international team before you ask anyone for money.
If your sector is regulated, check licensing first. Some services cannot legally be sold without registration or a licence. Confirm this before you present any offer.
The First Customer Test
The test runs over roughly two weeks: ten interviews, one written offer, one ask, one decision. The interview method follows The Mom Test's core rule, which is to ask about the customer's life and past behaviour, never about your idea. People lie politely about the future. They rarely lie about what they did last Tuesday.
Step 1: Define the problem in one costed sentence
Write: "For [group], the problem is [problem], which costs them [amount of time or money] every [week or month]."
"Small businesses need better marketing" fails this test. "Self-employed electricians in Reading lose five to six evening jobs a month because they cannot send quotes the same day" passes it. If you cannot fill in the cost, you have not found a problem worth ten conversations yet. Finding that out on day one is the test working.
Step 2: Recruit ten strangers, not ten supporters
Friends, family and fans will encourage you, and encouragement is noise. Recruit from trade associations, chambers of commerce, LinkedIn, sector forums and university societies. A workable message: "I'm researching how [group] handle [problem]. Could I have 15 minutes about the last time it happened?"
Guard against three sampling traps:
- All one place. Ten interviews from one city or one association tell you about that pocket, not the market.
- Over-helpful contacts. Some people love giving feedback and never buy. Ask what they last paid for in this area.
- Self-selection. People who respond fastest often have the strongest opinions. Note who ignored you, too; silence is data.
Step 3: Run the interviews about the past
Four questions carry the interview:
- When did you last deal with this?
- What did you do, step by step?
- What did it cost in time, money or stress?
- What have you already tried, and what happened?
Never ask "Would you buy this?" Log every answer:
| Ref | Role | Problem in their words | Last occurred | Cost to them | Already tried |
|---|
By interview six or seven, the same problems should repeat, in similar words. Those repeated phrases become your offer copy. If nothing repeats, either the group is too broad or the pain is too mild to pay for. Both findings end the test early and cheaply.
Step 4: Price the offer against the customer's own numbers
Write a one-page offer with four parts: the problem, the outcome, the price, the timeframe.
Anchor the price to the cost figures from your log. If your electrician interviews showed six lost hours a week at £30 an hour, the problem costs about £180 a week, or £700-plus a month. A £150-a-month quoting service prices well under the pain. A £400-a-month one has to argue harder. These numbers are illustrative; the method of using the customer's own arithmetic is the point.
To test willingness to pay more precisely, ask the four Van Westendorp questions: at what price would this be suspiciously cheap, a bargain, getting expensive, and too expensive to consider? The overlap across ten answers gives a rough range, not a final price.
Do not build anything yet. A spreadsheet, a phone number or a manual service tests demand as well as software does.
Step 5: Present, and treat objections as the product roadmap
Show the offer to your interviewees and ask three questions: Would you use this? What would stop you? What would you change?
Common objections and what to do with them:
- "We already use someone." Ask what would make them switch. Switching costs are part of your real market.
- "Too expensive." Ask what they would expect to pay, and what proof they would need first.
- "I need to think about it." Ask what specifically, and when they will decide. No answer usually means no.
Behaviour outranks words. Someone asking about start dates or payment terms is warmer than someone praising the concept.
Step 6: Ask for a commitment, scaled to what the law allows
Commitments run from weakest to strongest: a reply to a follow-up, a booked call with the decision-maker, a signed letter of intent, a deposit, a paid pilot. A booked call with the person who controls the budget is real evidence. A deposit is better.
If you take money: be clear on terms, refunds and delivery, because the Consumer Rights Act 2015 governs consumer contracts. If you are on a Student visa, stop at the letter-of-intent rung, and remember it evidences demand without making trading lawful.
Step 7: Decide against a threshold you wrote down before you started
Set the bar before the results come in, in writing, or you will move it. This is a preliminary experiment, not a benchmark; ten conversations cannot prove a market. Read the outcomes like this:
- Several strong commitments from the target group: run a small paid pilot and measure delivery costs as you go.
- Interest plus consistent objections, no commitments: change one variable, price, format or promise, and test again with ten new people.
- Commitments from a group you did not target: follow the evidence and redefine the segment. This happens more often than founders expect.
- Only polite interest: before concluding anything, check whether your questions were leading and your sample biased, then re-test with strangers.
- Nothing after a fair, unbiased test: stop or pivot, and keep the log. Walking away after two weeks is a success compared with walking away after a year.
Telling demand from politeness
| Usually genuine demand | Usually polite interest |
|---|---|
| Describes the problem in detail unprompted | Says the idea sounds interesting |
| Names the cost of the problem | Says they would "definitely" use it |
| Has already tried to fix it | Asks to be kept informed |
| Names a budget or a date | Offers general encouragement |
| Introduces the decision-maker | Promises to tell friends |
| Commits something of their own | Gives no next step |
One strong signal is a lead. A pattern across strangers is the beginning of evidence.
Using AI without fooling yourself
AI speeds up the analysis if you keep it honest:
- Transcription turns recordings into searchable text. Check quotes against the audio before using them.
- Tagging works when you supply the tags, such as "problem," "cost," "workaround," "objection," and spot-check a sample. Left to invent themes, models flatter patterns that are not there.
- Objection counts should be verified against the raw log before you act on them.
- Offer drafts in the customer's own words are a good AI use, tested on real people afterwards.
Privacy is not optional. If notes identify people, UK GDPR applies: get consent to record, strip names before uploading to any tool, and check the tool's data terms. The ICO publishes the relevant guidance. And no model can tell you what someone will pay; only the ask in step 6 does that.
A paying customer is not yet a business
The panel data's deeper warning is that early wins fade. Three numbers tell you whether commitments can become a company:
- Customer acquisition cost (CAC): sales and marketing spend divided by new paying customers.
- Monthly gross margin per customer: the fee minus direct delivery costs.
- Payback period: CAC divided by that margin.
Illustratively: a £300-a-month service at 60% margin leaves £180 a month. At £600 CAC, payback is about 3.3 months, workable. At £1,500 CAC, it is 8.3 months, and the same three pilot customers now describe a much weaker business. Track these from the pilot onwards, along with whether customers renew. Retention is the evidence the panel firms that sustained growth would recognise.
After the test
The test ends with one of four moves: pilot, refine, re-segment, or stop. Whichever you choose, keep the log. It is your evidence base for investors, for endorsement applications if you later take the founder visa route, and for your own next idea.
If the pilot holds and the work outgrows you, our guide on how to find a co-founder covers the search, and founder-market fit explains why your link to the problem matters to backers. For market context, see the UK startup statistics guide, and if you are building with AI, the early-stage AI accelerators roundup lists support options.
Ten conversations, one offer, one ask. Start recruiting this week. The government's own data says forecasts miss half the time; strangers who commit are the only forecast that does not.
