Friends like the prototype. The landing page looks polished and the feature list keeps growing. None of those responses proves that a customer has a problem serious enough to pay for a solution. Before committing months to development, a founder can investigate how people solve the task today, what the workaround costs them and whether they will make a meaningful commitment to a better alternative. The strongest early evidence often comes from behavior, not compliments.
The explanation is often more mundane than the dramatic phrase startup failure suggests. People may like the concept without experiencing the problem often enough to pay. They may already have a workable routine. The person who uses the software may not control the budget, or moving away from the existing system may be more difficult than expected.
A useful idea needs evidence from the situation in which it would operate. That evidence can be collected long before a polished product exists.
Begin with the last time the problem happened
Suppose the proposed tool helps small agencies coordinate property cleaning. It is tempting to ask owners, "Would you like a dashboard that organizes your cleaners?" Most people can imagine a benefit. Their answer says little about how urgently they need it.
A better conversation asks about yesterday's work. When did a cleaner last arrive at the wrong apartment? How did the agency discover the mistake? What information was missing? How much time did it take to fix? Who had to stop their own work to help?
The same principle applies to almost any business product. Ask about specific recent events, not the respondent's opinion of an imagined future. Concrete events are easier to examine. They expose existing tools, exceptions and the real cost of mistakes.
Y Combinator's Startup Library lecture on building products and talking to users emphasizes direct contact with people using the product. That advice matters because a founder's plan is usually less detailed than the customer's actual workflow.
Pay attention to the language people use when describing an inconvenience. Do they call it a minor annoyance, an embarrassing mistake or a serious financial loss? Does the problem occur every day or twice a year? A small annoyance repeated hundreds of times may be more valuable to solve than a severe but extremely rare problem.
Look for the workaround that already costs money
People generally attempt to solve problems before a startup arrives. They use spreadsheets, group chats, consultants, extra employees, established software or manual routines. Those workarounds reveal both the pain and the limits of a proposed solution.
If a company employs someone full-time to reconcile schedules, there may be an opportunity to reduce wasted effort. If the entire problem is handled with a two-minute weekly check, replacing that routine with a paid system will require a stronger reason.
Ask what the current process costs. Include the obvious invoices and the less visible consequences: missed sales, staff time, customer complaints and delayed decisions. Avoid presenting rough interview estimates as audited financial facts. The aim is to understand which activities customers already consider worth paying to manage.
Competition is useful evidence. A market with established products can show that buyers already spend money on the category. It also raises the burden of explaining why they would switch. The US Small Business Administration's planning and competitive-analysis guidance recommends investigating direct and indirect competitors, strengths, market segments and barriers to entry.
The strongest competitor may be inertia. A business can dislike its current software and still refuse to migrate because historical records, staff training and integrations are difficult to replace.
Build an evidence ladder, not a feature backlog
A founder can make decisions more deliberately by treating each new observation as a rung on an evidence ladder.
At the lowest rung, a respondent says the idea sounds interesting. That is a useful opening but weak evidence of buying behavior. Higher up, someone describes a recent, costly incident and shows how they tried to resolve it. Stronger still, a buyer agrees to allocate time to a trial or introduces the person responsible for purchasing. A paid pilot or a clearly documented purchase commitment supplies more evidence than a compliment, although even payment does not guarantee long-term retention.
The rungs are an original editorial framework, not an official certification method. Their purpose is to prevent a founder from treating several friendly conversations as proof of demand.
A team should record disconfirming evidence too. If most interviewees never encounter the problem, that observation deserves space in the notes. If one prospect loves a feature but lacks purchasing authority, record the limitation. A founder who only saves enthusiastic quotes is collecting marketing material, not conducting market research.
Set a decision rule before beginning. For example, agree to examine a defined number of target customers and continue only if several can document a frequent problem and a credible willingness to commit money or time. The exact thresholds are business choices, not universal startup laws.
Run a concierge experiment
Software is not always necessary to test the value of a service. If the idea is to send accurate cleaning schedules to agencies, offer to organize one small agency's week manually using existing tools. Charge a modest amount if the commercial relationship makes that appropriate, and explain what is manual.
The experiment tests the outcome customers want. Does the agency use the schedule? Do staff follow it? Is the information available at the moment a cleaner needs it? Are owners relieved, or do they keep using the old group chat because it remains more convenient?
A concierge test also exposes the features a first product actually needs. The founder may discover that the hardest part is capturing last-minute changes from staff, not presenting a sophisticated calendar. That changes what should be built.
Y Combinator's essential startup advice favors launching useful products quickly, speaking directly with users and iterating. A manual pilot can sit before a software launch when the hypothesis is about service value rather than technical feasibility.
Such tests have limitations. A founder's personal attention may produce a level of service that a scalable product cannot maintain. Manual costs may be hidden by unpaid effort. Be explicit about those constraints before treating a successful pilot as evidence of a profitable software business.
Ask for a commitment that resembles the future sale
A positive interview is not a purchase. To narrow the gap, invite the buyer to take an action that involves a genuine cost: schedule a pilot, supply non-sensitive example data, introduce the budget holder or agree on proposed commercial terms.
The action should be appropriate and transparent. Do not collect money while pretending the full product exists when it does not. If a refundable deposit is offered, explain the conditions in writing and follow applicable consumer and contract rules. Often a paid manual pilot is a clearer test than a speculative pre-order.
The sequence matters. First understand the problem, then demonstrate how a proposed workflow solves one part of it, and only then discuss price. People asked to name an acceptable price for a vague concept often invent numbers. A buyer who has experienced a working solution can give more meaningful feedback.
A prospect declining to pay is valuable information when the reason is understood. Perhaps the problem is not serious enough. Perhaps the solution lacks an integration that makes switching possible. Perhaps purchasing happens annually. Each explanation suggests a different experiment; none automatically proves the founder should add another ten features.
Include the person who signs the contract
In business-to-business products, the person using a tool and the person paying for it can have conflicting priorities. A receptionist may want faster scheduling; a manager may care about errors and reporting; a finance department may require vendor approvals; an IT team may object to a new application with broad data permissions.
A founder should map that buying process while the product is still flexible. Who experiences the pain? Who approves spending? Who can stop adoption? What training and data migration would be required? Which security or compliance requirements are mandatory?
The answers are part of product-market fit. A delightful interface that cannot pass procurement or migrate the customer's data may never be used.
For very small businesses, one owner may play all those roles. Even then, the owner might need a family member, accountant or office manager to approve operational changes. Look for the actual decision process rather than assuming the loudest enthusiastic respondent is the buyer.
Measure the pilot as a business process
Before running a pilot, write down one result the customer already cares about. It might be fewer missed appointments, less time spent preparing weekly reports or faster handling of customer requests.
Capture the starting point, the pilot's duration and the method for judging success. Include the work required to deliver the result. If a founder spends ten hours a week saving a customer two hours, there may still be learning value, but the current economics are weak.
Avoid interpreting every positive comment as a success. A customer may appreciate the attention while remaining unwilling to change routines. Retention and repeated use, particularly without constant reminders, can be more revealing than enthusiasm in a demonstration.
After a trial, ask what the customer would do if the service disappeared. Would they return to their previous method immediately? Would they make a budget request? Would a particular employee be affected? These answers help separate pleasant extras from operational dependencies.
A two-week research schedule
A small founder-led effort can start with a two-week sequence, provided the founder can reach the target audience. Spend the first days defining a narrow customer group and recording the current alternatives. Use the next several days for conversations about actual past events.
Then choose one repeated, costly problem and create the smallest possible test of its outcome. Show the result to a few relevant buyers and ask for a real next step. End with a decision memo: what evidence supports the idea, what contradicts it, which assumptions remain untested and what would justify further investment.
The timeline is an illustration, not proof that a particular business can validate demand in fourteen days. Complex markets, regulated services and long sales cycles require longer experiments. A founder's job is to reduce uncertainty at a reasonable cost, not to force reality into a calendar.
The evidence may support building a product. It may support a simpler service, a different audience or abandoning the concept. All are better outcomes than spending six months designing features no one asked for.
References and background
Our explainers link to source documents so readers can explore the evidence and interpret it in context. Websites and policies may change after publication.
- Y Combinator, *Essential Startup Advice*www.ycombinator.com
- Y Combinator, *Building Product, Talking to Users, and Growing*www.ycombinator.com
- US Small Business Administration, *Plan Your Business*www.sba.gov