Validation in 2026 is not 30 customer interviews where everyone tells you it is a great idea. People are polite. Validation is a paid signal — money on the table or a signed commitment. Anything else is theater.
This guide gives you five experiments you can run for under $1,000 each, in two weeks each, before you write a line of production code. The idea that survives all five is real. Most do not.
What changed in 2026
- AI made building cheap. Validation is the new bottleneck. Two engineers can ship anything in a weekend. The hard part is whether anyone wants it.
- B2B buyers got pickier. Procurement tightened, pilot budgets shrank. Validation has to be sharper than in 2021.
- Landing-page-only smoke tests still work. The conversion benchmarks have not moved much.
How to run a validation experiment
- One falsifiable claim per experiment. "Designers will pay $50/month for X" — yes or no.
- A real audience. Not your network. Buyers from Reddit, niche Slacks, X.
- A real ask. Email signup, paid pre-order, scheduled call.
- A real budget cap. $500–$1,000 per experiment. Past that, you are over-investing.
- A real timeline. Two weeks. If it takes longer, the experiment is wrong.
1. The smoke-test landing page
Build a one-page site describing the product as if it exists. Add a "buy now" or "join waitlist" button. Run $300 in targeted ads to a narrow audience. Measure click-to-signup rate and signup-to-paid intent.
Benchmark: 5%+ signup rate is interesting. 1%+ payment intent (even if you refund) is real. Below those, the headline or the audience is wrong.
2. The concierge MVP
Do the work manually for one customer. They pay you (real money, even $50). You deliver via spreadsheets, email, and Zoom. The product does not exist; you are it.
If one customer says yes and pays, run it for ten. If you cannot find ten, the market is too small or you are not the right person to serve it.
3. The paid pilot for B2B
For B2B ideas: get one company to sign a $5–25k pilot agreement before you build. Two pages, signed by procurement. Even a small commitment proves the buyer is real.
This is the single highest-signal validation. If you cannot get one paid pilot, you do not have a B2B startup yet.
4. The wizard-of-oz prototype
A working interface with a human behind the curtain. The user uploads a file; you process it manually and return results. They never know. Used by Zappos in the early days. Still works.
The catch: you cannot scale past 10–20 users. That is the point — it forces the validation question without the build.
5. The pre-sell campaign
Take 100 dollars on day one for something that ships in 90 days. Used heavily on Indiegogo and Kickstarter, but works equally well via a Stripe payment link sent to a niche email list. If 10 people pre-order, the demand is real.
The risk: you owe them a product. Set a refund policy and a clear ship date.
Comparison: validation experiments in April 2026
| Experiment |
Cost |
Time |
Signal strength |
| Smoke-test landing page |
$300–$1,000 |
1–2 weeks |
Medium |
| Concierge MVP |
$0–$500 |
2–4 weeks |
Very high |
| Paid B2B pilot |
$0 |
4–8 weeks |
Very high |
| Wizard-of-oz |
$200–$500 |
2 weeks |
High |
| Pre-sell campaign |
$200–$1,000 |
2–4 weeks |
High |
| Customer interviews only |
$0 |
4 weeks |
Low |
Common mistakes to avoid
Asking "would you pay for this?" Everyone says yes. The answer is meaningless. Replace it with "would you pay $X right now?" and put a payment link in the message.
Validating with friends and family. They are biased. Ten cold strangers from a niche Slack are worth more than a hundred warm referrals.
Building before validating. "I will just ship the MVP and see" wastes 3 months. The smoke test answers the same question in 2 weeks.
FAQ
What if my idea is too novel for a smoke test?
Then the concierge MVP. If a service version cannot get paying customers, the product version will not either.
How many "no"s before I kill the idea?
Ten paying-customer attempts. If you cannot find ten people willing to pre-pay, refine the idea or move on.
Is YC right that validation is overrated?
Their version of "build first" works because they fund founders who will iterate fast. For most founders without that backing, validation first is cheaper.
Where to go next
For related guides see How to ship a SaaS in 30 days in 2026, How to find a co-founder in 2026, and Cost of running a side project in 2026.