Validate your startup idea before you build it
How do you validate a startup idea?
One experiment at a time, written down before you run it. Six boxes, one sentence each. There's a made-up example in there already: swap in your own idea and watch the sentence at the bottom write itself.
Everything stays in your browser: nothing is sent or stored. Reload the page and the example comes back.
We believe…
A sentence that could turn out to be wrong. If it can't fail, it isn't a hypothesis.
We'll test it with…
People who have the problem today, not people who might have it one day.
through…
The cheapest test that answers the question. The six most common ones are below.
measuring…
An action that costs the person something: time, data, money.
We go ahead if we reach…
Set it now. Moving it once you've seen the numbers is the commonest way to fool yourself.
within…
A test with no deadline never ends, so it never decides anything.
Your experiment in one sentence
We believe neighbourhood restaurants waste food because they order by gut feel, and would pay for an order forecast. We'll test it with owners of independent restaurants who place their own orders through a landing page with the promise and a 'Book a demo' button, measuring demo requests as a share of visitors. If we reach 15 requests, at least 5 from people we don't know within three weeks of the page going live, we go ahead; if not, we change the hypothesis.
Which tests should you use to validate an idea?
The cheapest one that answers your riskiest question. It's worth the effort: according to CB Insights (2026, 431 venture-backed startups shut down since 2023, mostly in the US), 43% named poor product-market fit among the reasons.
| Test | What it proves | What it doesn't prove | Cost | Time |
|---|---|---|---|---|
| Interviews | That the problem is real, and how customers describe it in their own words. | That they'll pay: people are polite about other people's ideas. | Low | Medium |
| Landing page | That the promise pulls people in: clicks, sign-ups, demo requests. | That the finished product will keep them. | Low | Low |
| Smoke test | That someone goes through the motions of buying: sees the price, clicks 'Buy', learns it isn't ready yet. | That they'll still be there after the first month. | Medium | Low |
| Concierge | That the outcome is worth having: you deliver it by hand to a few real customers. | That it can be automated with healthy margins. | Medium | High |
| Wizard of Oz | That the experience works: users see software, and you're behind the curtain. | That the technology is feasible, or affordable. | High | Medium |
| Clickable prototype | That people understand the flow and can find their way through it. | That they'll actually use it, week after week. | Medium | Low |
Cost and time are relative to one another across these six tests: they depend on your market, your channel and how close you already are to your customers.
When should you build the MVP?
When the cheap tests have answered 'do they want it?' and the question left is 'will they really use it, and come back?'. Only a product can answer that one.
It's time if
- you beat the pass mark you set beforehand, not one you picked afterwards;
- someone has already put in time or money: a pre-order, a pilot, a letter of intent;
- you can name the one feature that has to work brilliantly on its own;
- the concierge can't keep up: too much manual work for the customers you have.
Not yet if
- the only yeses come from friends and colleagues;
- you moved the pass mark after seeing the numbers;
- you want to build it to find out whether anyone cares: a test is cheaper.
What we do at this stage
This is where we start with founders: interviews, test landing pages, clickable prototypes and the tracking to read the numbers honestly. Once the pass mark is beaten, we move on to MVP development and the pre-seed stage. Every stop is on the map of startup stages.
Questions about validation
How many interviews do you need to validate an idea?
There's no magic number: you stop when the answers start repeating and you're no longer learning anything new. Who you talk to matters more: people who have the problem now, not kind friends. Ask what they did last time, not what they would do.
How do I validate a startup idea without spending money?
Start with interviews and a landing page with a form: they cost mostly time. Reach the right people by hand, as Paul Graham argues in 'Do things that don't scale' (2013). Budget comes later, when you want more traffic for a smoke test.
Is a landing page with a waiting list enough?
It's a decent first signal, but an email address costs the person very little. Make the action more committing: a booked demo, a pre-order, a pilot. The more it costs the customer, the more the signal is worth.
How do I know if I have product-market fit?
A widely used method is the 40% test: ask users how they would feel if the product disappeared; below 40% 'very disappointed', companies usually struggle. First Round tells how Superhuman went from 22% to 58% by working on that number. It's a single case, but the method travels well.
Can I validate an idea while I'm still employed?
Yes: interviews, landing pages and smoke tests fit into evenings and weekends. Concierge and Wizard of Oz tests take more time, so save them for when the first signals look good. What you can't put off is setting the metric and the pass mark before you start.
Further reading
Got an assumption to put to the test?
We'll help you pick the cheapest test that answers your question.
Let's talk