Insights / startup-foundations

Does Your Interview Evidence Justify Building an MVP?

By Daniel Tonkopiy | 2026-10-26

Interview evidence justifies building an MVP only when buying behavior, not agreement, clears thresholds set in advance. Founders at the idea-to-validation stage should fix an observation window, then count accounts with a recent purchase for the problem, a costed workaround and a budget owner who completed a paid or dated next step. Build for those accounts if the thresholds hold; change one variable or stop if they do not.

The outcome and observation window

This guide is for founders at the idea-to-validation stage who have run customer discovery interviews and now have to decide whether the evidence is strong enough to spend money and months on a minimum viable product. The question is not whether the interviews went well. It is whether the behavior of the people who would pay has moved far enough to justify building. If the interviews are still ahead, start with the customer discovery questions that reveal buying intent. If buyers already said yes and then went quiet, the diagnosis in when customer interviews say yes but nobody buys comes first.

Treat the interviews as a measurement, and write down three things before the next conversation. The outcome is the behavior that would justify an MVP, defined as a completed step by the person who controls the budget, such as a paid pilot, a deposit or a signed order with a start date. The observation window is a fixed number of conversations or weeks after which the count is final. The thresholds are the counts that mean continue, change or stop. Writing them down in advance keeps a run of friendly conversations from quietly lowering the bar.

A federal training program built around customer discovery uses the same idea of a fixed endpoint. The National Science Foundation expects each team in its national Innovation Corps program to have conducted at least 100 interviews with potential customers and other stakeholders by the end of the program. In testimony to a House subcommittee in December 2017, the NSF's Assistant Director for Engineering said that teams reaching a No Go decision at the end of the curriculum are of equal benefit to teams reaching a Go decision. A program of that size is not a target for a single founder, and the number of interviews is an input, not the result. The part to borrow is the fixed endpoint and the acceptance that stopping is a valid outcome.

Recent purchasing behavior: definitions and evidence

In this framework, recent purchasing behavior is the most direct sign that a problem already has a budget. Define it before the window opens: the account has spent money or paid staff time on this specific problem within a lookback period that fits how often the problem occurs, such as the past year for an annual purchase or the past quarter for a weekly one. Spending on a neighboring problem does not count.

Then grade the evidence with the said, seen, done and assumed tags from the customer interview notes template. A participant who mentions hiring a contractor last spring has given a said entry. An invoice, a purchase order, a subscription screen or a contractor the founder meets is a seen entry. Only seen entries count toward the threshold. The comparison that matters is between two counts: how many accounts describe the problem as expensive, and how many can show money leaving the business because of it. When the first count is far larger than the second, the interviews have measured concern, not a budget.

Current workaround: comparison and caveats

An account that has the problem usually already does something about it, even if that means tolerating it. The MVP will be compared with that workaround, not with nothing. Record what the workaround costs in money and hours, who maintains it, how often it fails and what a failure costs. Ask to see it working rather than hear it described.

The Small Business Administration's market research guide lists in-depth interviews among direct research methods and, among the questions to answer about a market, asks what potential customers pay for the alternatives. For an MVP decision, turn that question into a comparison for each account: the observed cost of the workaround over a period against the price the founder intends to charge over the same period.

Three caveats keep the comparison honest. A workaround cost that the founder estimated is an assumed entry, not evidence, until the account confirms it from its own records. A workaround that is cheap but painful can still win, because it is already approved and familiar. And a workaround whose cost is spread across several people may be real but invisible to the budget owner, which makes the next indicator decisive.

Decision-maker and budget owner: scenarios and review triggers

Buying behavior counts only when it comes from someone who can authorize the purchase. Most early accounts fall into one of three scenarios.

• The user is the budget owner. This is often the case in owner-run businesses. A completed step from this person counts directly.
• The user is not the budget owner, and the owner took part. Count the step only if the owner made or approved it, not if the user promised to raise it internally.
• The budget owner did not appear. The account's evidence describes the problem, not a purchase. It does not count toward the build threshold, however strong the interview was.

Set review triggers as well: conditions that pause the count before the window ends because the measurement itself may be broken. Useful triggers include every completed step coming from accounts introduced by friends or investors, a discount offered to secure a commitment, a commitment from someone who turns out not to hold the budget, and evidence concentrating in a subsegment that was not part of the original hypothesis. A trigger does not end the test. It means checking whether the count still measures the outcome defined at the start. How to reach the approver without selling is covered in how to run a customer interview without pitching.

The interview evidence map for an MVP decision

The map below compares, indicator by indicator, what the interviews found with what buyers did. Fill it in for every account at the end of the window, then total each row. It is a measurement version of the evidence map used in earlier guides: instead of choosing accounts or diagnosing a stalled yes, it decides whether the count justifies building.

Read the committed next step row last and give it the most weight. The other rows explain why the count came out as it did. Only the committed next step measures the outcome.

A hypothetical worked example from greenhouse growers

The following is a hypothetical example with invented details. It illustrates the method and does not describe a real company, customer or measured result.

A founder wants to build sensors and an app that warn small commercial greenhouses when a drip irrigation line clogs, before a bench of seedlings dries out. Before more interviews, the founder writes down the outcome, a window of eight weeks and fourteen interviews, and three thresholds. Continue to an MVP if at least four accounts make a paid pilot commitment approved by the person who signs for equipment. Change one variable if two or three do. Stop if one or none does.

At one nursery, the founder arrives before opening and watches the head grower walk the propagation benches, pressing a fingertip into the soil of tray after tray to find the dry ones, then clearing a blocked emitter with a pin. The grower says this check takes most of the first hour every morning and that a missed clog last summer cost a full bench of young plants. The founder asks to see the record of that loss and is shown the replacement order.

At the end of the window the map reads as follows. Eleven of the fourteen growers say they would use the product. Seven show a record of trays lost to irrigation failures, and five show a purchase in the past year aimed at the problem, such as replacement emitters or a technician's call-out. The person who approves equipment spending took part in six conversations. Three accounts agree to a paid four-week pilot with a deposit and a start date, each approved by the owner.

Three is below the continue threshold, so the founder does not build the full product. The pattern row explains the result: all three commitments come from nurseries that raise young plants from seed under mist, where in this example a dry hour can ruin a bench, while the growers of finished plants who said yes have a workaround that costs them little. The founder makes one change, narrowing the segment to propagation nurseries, keeps the same product, price and thresholds, and opens a new window with new accounts. The three pilots go ahead as a manual service with off-the-shelf sensors, so that the paying accounts get value while the decision is retested.

The assumptions are written down with the decision: that the owners who approved the pilots will pay the full price after the trial, that the replacement orders the growers showed reflect typical losses rather than one bad season, and that a manual service is close enough to the planned product to teach something about it.

Decide what to continue, change, or stop

Close the window on the date set at the start and apply the thresholds as written. Each outcome has a specific next action.

• Continue. The committed next steps reached the build threshold. Build the smallest product that delivers what those accounts paid for, and nothing for accounts that only said yes. The commitments define the first scope.
• Change. The count fell between the thresholds, or the steps clustered in one segment. Change one variable, such as the segment, the buyer or the offer, and run a new window with new accounts and the same thresholds, so that the results can be compared.
• Stop. Few or no budget owners completed a step, the workaround costs less than the intended price, or recent spending is absent. Record what was learned and stop spending on this hypothesis. A stop reached on evidence is a result, not a failure of the process.

Write the decision down with the date, the counts in each row of the map and the assumptions behind it. The pillar guide on how to validate a startup idea before building covers the broader build, narrow or stop decision and the smallest honest offer that produces a commitment to count.

Common mistakes and limits

• Moving the thresholds after seeing the data. If the count misses, the honest options are change or stop, not a lower bar.
• Extending the window to rescue the hypothesis. More conversations of the same kind rarely change a count that has stalled.
• Counting interviews instead of accounts. Several conversations at one account are one data point.
• Counting free commitments. A free pilot measures willingness to try, not to pay. If one is offered, keep it out of the build threshold.
• Building for everyone who said yes. The MVP scope should follow the accounts that committed.

The method has limits. It suits business sales with an identifiable approver better than consumer products, where usage and purchase data say more than interviews. With a small number of accounts, any threshold is a judgment and any result is a hypothesis for the next window, not a statistical finding. The indicators, thresholds and evidence map are an editorial framework, not a measured result.

Related guides and next step

• Questions about purchases, workarounds and approvers: Customer Discovery Questions That Reveal Buying Intent
• Recording each interview with evidence tags: Customer Interview Notes Template for Founders
• Diagnosing positive interviews without purchases: When Customer Interviews Say Yes but Nobody Buys
• The broader build, narrow or stop decision: Validate a Startup Idea Before Building

Daniel's go-to-market mentoring page lists defining a pilot outcome, a buyer commitment and the decision that follows the pilot, and choosing what to record each week, including conversations, agreed next steps, paid pilots and reasons deals stop, among the work a session covers. That description is self-reported on his site.

Bring your evidence map and thresholds to a free 30-minute strategy session to review your customer discovery interviews decision with your actual evidence and constraints.

Frequently asked questions

Does your interview evidence justify building an MVP?

It does when buying behavior clears thresholds set before the observation window: enough accounts with a seen recent purchase for the problem, a workaround that costs at least the intended price, and a budget owner who completed a paid or dated next step. Agreement and enthusiasm in interviews do not count toward the threshold on their own.

What evidence should a founder collect for customer discovery interviews?

For each account, record the last purchase made for the problem and whether it was seen, the current workaround with its cost from the account's own records, who approves spending and whether that person took part, and every step completed after the conversation. Tag each entry as said, seen, done or assumed, and count only seen and done entries toward a build decision.

What are common mistakes in customer discovery interviews?

Treating agreement as a commitment, counting conversations instead of accounts, interviewing users who cannot approve spending, accepting estimated workaround costs as evidence, and lowering the bar or extending the window after the results come in.

How long should the observation window be before an MVP decision?

Long enough for the buying step being measured to happen at least once in a normal sales cycle for the segment, and fixed before it starts. Define it as a number of conversations, a number of weeks or both, then close it on the planned date whatever the count shows.

Sources

1. Information for Accepted National Teams, NSF Innovation Corps, U.S. National Science Foundation Checked 2026-10-26
2. Testimony of Dawn Tilbury, NSF Assistant Director for Engineering, From Lab to Market: A Review of NSF Innovation Corps Checked 2026-10-26
3. Market research and competitive analysis (Plan your business guide), U.S. Small Business Administration Checked 2026-10-26
4. Go-to-Market Mentor and GTM Strategy Advisor Checked 2026-10-26

Customer discovery | MVP decision | Startup foundations | Founder playbook