Short version
Week one: write assumptions, define the segment, build a script, recruit. Week two: run ten interviews, score them, then ask three people for a small paid commitment. Decide on day fourteen against thresholds you set on day one.
Day 1: choose the decision and write your stop rules
Two weeks can give you enough evidence to choose a next investment, not prove that a business will succeed. The sprint works when you test one customer segment, keep the offer small, and decide against rules written before anyone compliments your idea. Avoid spending the fortnight making something impressive to show.
Use a hypothetical example: a manual booking-support service for independent physio clinics whose receptionist recently left. Your eventual idea might involve software, but your first question is whether owners have an unresolved problem they will pay someone to handle. Keep patient information out of discovery and define safe boundaries before any delivery.
List the beliefs the idea depends on, then rank them by how damaging it would be if they were false. Put the top three into the template below. Your thresholds are practical decision rules for this sprint, not universal validation standards. Choose them before outreach and do not quietly lower them later.
- Assumption: [specific belief]. Evidence needed: [observable behaviour]. Test: [conversation or offer]. Decision threshold: [result required by day fourteen].
- Example problem rule: among ten qualified conversations, at least six describe a recent unresolved booking problem and an active workaround. These are illustrative targets, not research benchmarks.
- Example commitment rule: offer a clearly scoped paid manual pilot to three qualified buyers; require two paid acceptances before investing in software. If purchasing cannot finish in two weeks, record that limitation rather than treating a verbal yes as payment.
- Stop rule: if the problem threshold fails, stop this segment test. If pain exists but the offer fails, revise the offer or segment before building.
Free tool: Startup idea validation checklistA 21-step checklist that takes an idea from assumption to go-to-market: write and rank assumptions, interview ten customers about past behaviour, ask for money, measure product-market fit, then turn the evidence into positioning and a first channel.Run it on your idea
Day 2: turn assumptions into an evidence plan
Give each risky assumption a row in an evidence table. Include what would support it, what would contradict it, where you will look, and its current status. Use supported, contradicted, and unknown rather than a single optimistic score that hides gaps.
For the clinic service, 'owners lose valuable time' needs accounts of recent interruptions and the work displaced. 'Owners can approve a pilot' needs an explanation of the purchasing process. 'We can deliver the outcome' needs a feasible workflow, including permission and privacy requirements, not just expressions of customer interest.
Block out the rest of the fortnight now. Reserve interview slots on days five through nine, follow-up time on days ten and eleven, and commitment conversations on days twelve and thirteen. If you work full-time, decide whether you can protect those hours; otherwise use the same sequence across three weeks rather than rushing the evidence.
Create one notes format today: participant identifier, segment fit, recent incident, workaround, cost, decision roles, supporting evidence, contradictions, and next action. Consistent notes make different calls comparable. Store contact details separately where practical, and ask permission before recording anything.
Day 3: define a segment you can actually screen
Turn the customer assumption into a profile with a trigger, current behaviour, and explicit disqualifiers. 'Small clinics' is not enough. 'Independent clinics where a recent admin departure left practitioners managing bookings' gives you something concrete to check.
Identify who feels the problem and who approves spending. They may be different people. A practitioner can explain the workflow while an owner explains the budget and timing. Keep your first batch within one operating situation so patterns do not disappear inside unrelated stories.
Write a screener: 'Has an admin staffing change affected how you handle bookings recently? When did it happen, and who handles the work now?' Do not tell people which answer qualifies them. Exclude situations your first offer cannot serve, such as purchasing controlled by a parent organisation with a long approval process.
Use the ICP builder if it helps keep the profile concise. Your output today is a profile you can read aloud, a neutral screener, and a list of possible recruitment sources. Add unknowns rather than decorating the profile with invented customer facts.
Free tool: Ideal customer profile (ICP) builderDescribe your idea and get three narrow, reachable customer segments with their profile, the pain they already pay to solve, where to find ten of them this week, and a disqualifier.Run it on your idea
Day 4: write the script and start recruiting
Build a script with five to eight backward-looking questions, linking each to an assumption. Read it aloud. If a question can be answered with 'Sure, sounds useful', replace it with a request for a real example. The interview question generator can help draft prompts; you still need to check their neutrality.
Start outreach today rather than waiting for a perfect script. Use two channels where the trigger is likely to appear, such as relevant professional communities and targeted introductions. Follow community rules and do not imply that you know a prospect has an internal problem.
Invite more people than you need because scheduling and screening will remove some. Twenty potential or provisional slots for ten completed calls is a planning cushion, not a predicted attendance rate. Avoid making twenty firm appointments you cannot honour; keep a reserve list and release unnecessary holds promptly.
- Interview opening: 'I am researching how clinics handle booking work after staffing changes. I would like to understand what happened, rather than show you a product.'
- Core prompts: 'Tell me about the last time this caused a problem. What did you do next? What else did you try? What time or money did that take? Who decided on that response? When will you review it?'
- Recruiting template: 'Have you dealt with [specific situation] recently? I am researching how people handle it. Could I ask about your experience for fifteen minutes? I can do [three concrete slots]. This is not a sales demo.'
Free tool: Customer interview question generatorPaste your startup idea and get eight non-leading customer discovery questions, each mapped to the assumption it tests, plus the leading questions to avoid. Free with a Vouch account.Run it on your idea
Day 5: run two calls and repair the process
Run your first two qualified interviews today. Open with the screener, explain how you will use the notes, and ask permission for any recording. Spend most of the conversation on one recent episode rather than collecting a broad catalogue of complaints.
After each call, score the evidence before the next one. Capture a brief factual account, any exact wording you recorded accurately, and your separate interpretation. If someone says the work takes ages, ask what happened on a particular day instead of translating that phrase into an invented cost.
Review the process after the second call. Were you interrupting, pitching, or asking several questions at once? Improve the wording without changing the underlying assumptions. Record meaningful script changes so you can recognise when later answers are not directly comparable.
Finish today by confirming the next appointments and asking suitable participants for one referral. A useful request is 'Who else has handled this recently?' rather than 'Who would like my idea?' Your running target is two completed qualified interviews, not two friendly conversations with anyone available.
Days 6 and 7: build the sample without broadening it
Day 6: run two more qualified interviews, bringing the target to four. Keep using the same core questions and record disconfirming details as carefully as supporting ones. Check whether the trigger actually changed behaviour or merely happened around the same time as an unrelated frustration.
Spend your recruiting block on the channel producing real matches, not merely replies. Track messages sent, replies, screened matches, booked calls, and completed calls. A reply rate below roughly one in ten is a prompt to investigate, not proof that the segment is wrong: message clarity, trust, timing, deliverability, and the contact source can all affect it.
Day 7: run two more calls, aiming for six completed interviews. Follow up with people who offered referrals, and fill gaps for days eight and nine. Keep the invitation short, offer three specific times, and make it easy to decline.
End day seven with a short review of recruitment quality. If you are missing the target, diagnose whether people are not responding, failing the screener, or missing appointments. Repair that specific problem. Do not silently admit unrelated customers just to make the interview count look healthy.
Days 8 and 9: finish ten interviews and test contradictions
Day 8: run two calls, aiming for eight completed interviews. Compare the latest accounts with your evidence table before each conversation. If several people describe the same issue, ask the next participant about a concrete episode without mentioning the emerging pattern; you want independent evidence, not agreement with your summary.
When three calls in a row tell a similar story, treat that as a sign of possible saturation on that assumption. Shift follow-up depth towards the next risky question, such as authority or timing, while keeping the core questions consistent. Repetition is useful, but three agreeing calls do not turn a small sample into certainty.
Day 9: complete the final two calls for a target of ten. Deliberately examine contradictions. Perhaps owners describe interruptions but are happy with a temporary workaround, or practitioners want change while owners see no financial consequence. Those differences may decide whether your offer has a buyer.
By tonight, every interview should be scored. If only seven qualified calls happened, report seven. Decide whether there is enough evidence to reject a specific assumption or whether the sprint remains inconclusive. Extending recruitment with a written purpose is better than relabelling unqualified chats as validation.
Day 10: separate problem evidence from solution enthusiasm
Review the complete evidence table, grouping accounts by assumption rather than by how enthusiastic the person sounded. Count independent participants, not repeated mentions from one particularly talkative interviewee. Include where each participant came from so a single referral network does not masquerade as broad agreement.
Weak evidence is a compliment, a hypothetical promise, or a general complaint with no recent example. Stronger problem evidence includes an active workaround, a recent search for help, spending, or a consequence the participant can explain. A customer-provided estimate is useful, but label it as an estimate rather than a measured result.
For a worked example, imagine seven of ten clinic owners describe a recent unresolved problem, five already spend time on a workaround, and three can approve a small service purchase. Those invented results would support offering a pilot to the three buyers. They would not establish that your software idea is wanted or that the economics work.
Write a paragraph answering: what problem repeats, for whom, after which trigger, and with what consequence? List the biggest remaining contradiction underneath. If the problem threshold has already failed, stop or redesign the segment test instead of forcing the remaining days into a sales exercise.
Day 11: design a paid test you can deliver honestly
Translate the repeated problem into a small manual service or concierge pilot. Sell a defined outcome, not a promise of future software. Be explicit about what will be manual, what is excluded, and what access or customer effort is required.
For the clinic example, the first offer might address a narrow administrative workflow without exposing sensitive patient information. Before promising delivery, check consent, permissions, and practical responsibilities. If you cannot serve the workflow safely, change the scope or choose a different test rather than treating those constraints as details for later.
Set a real price that would inform your business decision, and write the payment timing. Free trials can test usefulness, but they do not answer whether a customer will pay. A heavily discounted pilot tests that discounted offer; do not assume it proves demand at your intended price.
Vouch can help founders connect interview evidence to assumptions and produce an LOI or pilot agreement for a customer to sign. Whatever format you use, make sure both sides understand the scope and obligations. Get appropriate advice when the agreement or service creates legal or regulatory questions.
- Pilot template: For [specific customer], we will deliver [bounded outcome] during [dates], using [manual process].
- Terms template: The fee is [amount], due [date]. The customer provides [inputs and access]. We exclude [work outside scope]. Cancellation and refund terms are [clear conditions].
- Learning template: We will review [observable outcome] on [date]. A successful pilot would justify [specific next investment], not an unlimited product build.
Days 12 and 13: ask three buyers for a costly commitment
Day 12: contact the three qualified buyers whose problems match the offer. Keep this separate from the discovery interview. Say: 'You described [recent problem]. I can offer [scope] during [dates] for [price]. Would you like to proceed under these terms?' Give them room to say no.
Ask for the commitment you defined on day one: a prepaid pilot, a paid manual engagement, or a specific signed agreement where payment cannot yet happen. These signals are not equivalent. Money received provides stronger purchase evidence than an unsigned proposal; a nonbinding LOI records intent but does not guarantee revenue.
Day 13: follow up on open decisions and clarify blockers. Ask what prevents proceeding, who else must approve, and when that decision can occur. Record the actual answer. Do not rename a procurement delay as rejection, but do not count it as a sale either.
Preserve the meaning of the test. If nobody accepts, do not strip out the price or promise extensive custom work just to collect a yes. You may learn the scope, timing, trust, or customer group is wrong. Record any revised offer separately so tomorrow's decision compares like with like.
Day 14: decide, document, and schedule the next action
Compare the results with your original thresholds. If the hypothetical sprint produced seven qualifying problem accounts and two paid pilots from three offers, it met the example rules. The next step is delivering those pilots and measuring outcomes, not immediately building a full product.
If problem evidence is strong but no buyer commits, identify the most plausible unresolved assumption and design a bounded follow-up. You might change the offer or test an adjacent segment while keeping the problem. If the problem threshold fails, stop this direction unless you can name a concrete reason the test itself was invalid.
Use a third category when necessary: inconclusive. Missing interviews or a documented purchasing delay can justify a short extension with an owner, deadline, and exact evidence needed. They do not justify declaring success. Keep the original result visible alongside the extension.
Write and share the decision paragraph below. Then put the next action on the calendar: pilot kickoff, a revised segment's first screening call, or closing the experiment. The value of two weeks is not a validation badge. It is replacing a vague hope with a smaller, evidence-backed decision.
- Decision template: We tested [segment, trigger, and offer] between [dates]. Our thresholds were [rules]. We observed [counts and evidence], with [contradictions and limitations]. We will [proceed, revise, stop, or extend] because [reason]. [Owner] will complete [next action] by [date].
Questions founders ask
Is two weeks realistic alongside a full-time job?It can be, if you can protect time for recruiting, ten short calls, same-day notes, and follow-ups. Otherwise spread the same sequence across three weeks. Reduce unnecessary preparation, not screening quality or evidence review, and be honest about what you completed.Should I build a landing page during the sprint?Only if it helps recruit suitable people or test a specific message. A simple idea page with a waitlist can capture interest, but signups do not establish problem severity or willingness to pay. Keep page-building from displacing interviews and commitment conversations.What if I cannot get ten interviews?Inspect the recruitment funnel before lowering the target. Check whether people see the invitation, reply, qualify, book, and attend. Improve the message or source, and reconsider the segment if it remains hard to reach. Report the actual sample and extend only with a specific learning goal.Does a signed letter of intent count as validation?It is evidence of intent, with strength depending on the signer's authority, specificity, conditions, and obligations. A nonbinding LOI is not a sale or guaranteed revenue. Record it separately from payment and use it to establish the next concrete purchasing step.Can I use this plan for a consumer business?Yes, but adapt the commitment and delivery method. A transparent paid manual service or clearly described preorder may fit better than a B2B pilot. Explain availability, cancellation, and refund conditions, and test with the intended customers rather than relying on supportive friends.What if buyers need more than two weeks to approve a pilot?Use the sprint to test the problem and map the real approval process. Record budget-owner involvement, required reviews, dates, and any signed intent without pretending procurement is complete. Extend the purchase test with a deadline, or choose a smaller offer the buyer can legitimately approve.
This story was originally published on 22 July 2026 at https://vouch.specky.space/blog/validate-a-business-idea-in-two-weeks.
Top comments (0)