Identify the real customer
Ask what problem the startup solves, who pays for a solution and why that customer chooses it. Sometimes the end user and paying customer are different people.
Investment funding shows that investors backed a project under certain conditions. It does not prove recurring profit or that the model travels to another country.
Separate the product from distribution
A strong product can fail without a workable customer acquisition channel. Find out whether the company grows through referrals, search, partnerships, paid acquisition or long-term enterprise contracts.
Replicating the interface does not replicate distribution, reputation, access to data or accumulated know-how. Look for poorly served problems rather than an app to clone.
Look for constraints
Examine service costs per customer, seasonality, refund rates, implementation time and local legal requirements. For some models, third-party integrations are the largest barrier.
Label company claims as claims. Do not treat marketing estimates of market size as independently verified facts.
Make an independent hypothesis
Write down: “People in group X have problem Y, currently addressed through Z. We will test alternative W.” Then run your own experiment without copying another company’s brand or protected material.
A sound idea survives questions about paying customers, frequency of the problem, value delivered and distribution costs.