How to Test Demand for a Product

Testing demand means getting a real person to do something that costs them something, in response to a product that may not exist yet. Everything else is research.

That definition rules out most of what gets called demand testing. Surveys do not qualify. Neither do interviews, competitor analysis, or market-size spreadsheets. Those are all useful for deciding what to test — they simply cannot tell you whether anyone wants it, because none of them require the respondent to give anything up.

This page covers six methods that do qualify, what each one actually measures, and when to reach for which.

The six methods compared

The pattern is worth noticing: the strongest signals mostly come from methods that ask for money or for a commitment, not from methods that ask for attention.

Landing page and waitlist

The default. Build a page describing the product as if it ships today, give it one action, and count signups.

Measures: whether the pitch is interesting enough to trade an email for.

Watch for: waitlist signups are the weakest of the credible signals. An email costs almost nothing, and waitlist-to-customer conversion is frequently brutal. Treat a healthy waitlist as permission to run a stronger test, not as validation.

Preorder or deposit page

The same page, except the action is a payment or a refundable deposit.

Measures: whether the problem is worth money today. This is the highest-quality signal available before a product exists, by a wide margin — it is the only one where the respondent's answer costs them the same currency your business runs on.

Watch for: the ethics are real and non-negotiable. Taking money for something you have no concrete plan to deliver is fraud, not a growth tactic. Refundable deposits with a stated timeline, or authorisations you do not capture, get you most of the signal without that problem. Say plainly on the confirmation screen that it is not built yet.

Cold outreach to a real ICP

Instead of waiting for traffic, go and find the exact people described in your assumption and contact them directly.

Measures: demand within a defined segment — which is far more actionable than undifferentiated interest, because a reply rate is attached to a describable group of people.

Watch for: this is the method most sensitive to targeting. A poor reply rate from a badly-chosen list looks identical to a poor reply rate from a dead market, and founders regularly abandon good ideas on the strength of that confusion. Before concluding the market is silent, check whether you were talking to it. Follow the applicable rules — GDPR, CAN-SPAM, and their local equivalents — which generally permit B2B outreach with an opt-out and do not permit buying consumer lists.

Paid ads to a landing page

Buy traffic, measure what it does.

Measures: demand at a known cost of acquisition, which is the one number here that speaks directly to whether a business is possible.

Watch for: it is the most expensive way to be wrong, and results are heavily confounded by ad quality. A bad result may mean nobody wants the product or that your creative was weak, and separating those takes several rounds and real budget. Use it after a cheaper test has already shown signal, to find out what demand costs — not to find out whether it exists.

Concierge delivery

Deliver the outcome manually to a handful of people before automating anything. No product, just you doing the work by hand.

Measures: whether the outcome is valuable — the cleanest possible read, since nothing is obscured by the quality of an interface.

Watch for: it does not scale and it is not supposed to. The trap is enjoying it: manual delivery to five delighted customers can absorb months while teaching you nothing new after week three. Set an exit condition before you start.

Pilot with letters of intent

For B2B and enterprise, where a single deal is large and buying cycles are long.

Measures: institutional willingness to buy, which is different from individual enthusiasm and is the thing that actually matters in that market.

Watch for: a non-binding LOI is worth considerably less than the excitement it generates. A signed pilot with a budget line attached is real evidence; a warm letter from a champion with no procurement authority is a well-formatted opinion.

Choosing between them

Answer three questions.

How expensive is being wrong? If a wrong answer costs you a weekend, run the cheap test. If it costs a year, buy the expensive signal.

Do you know who your buyer is? If yes, cold outreach beats waiting for traffic — it is faster and the result is attached to a segment. If no, you have a targeting problem to solve first, and no demand test will fix it.

Is your risk demand or delivery? These methods all test demand. If you already know people want it and the open question is whether you can build it, you need a technical spike instead.

Reading the result honestly

Two rules do most of the work.

Set the threshold before you run. Write down the number that counts as a yes in advance. Post-hoc, every result looks encouraging, and this is not a character flaw — it is how everyone reads their own data.

Separate a bad idea from a bad test. A negative result has at least four possible causes: the market does not want it, you targeted the wrong people, the pitch was weak, or the sample was too small. Only the first is about the idea. Before abandoning anything, rule out the other three — and if you cannot, the honest verdict is inconclusive, not no.

An inconclusive result is not a failure. It is an instruction to run a better-aimed test.

How theLab runs this

theLab automates the third method and combines it with the first. It takes your idea, extracts the assumption and the ICP, generates the landing page and the outreach sequence, sources real contacts matching that profile, runs the sequence, and returns a verdict — Validated, Inconclusive, or Not Validated — based on signups and preorders rather than replies or sentiment.

The reason it reports Inconclusive as a distinct outcome is the rule above: a quiet result frequently means the aim was wrong rather than the idea, and collapsing those two into a single "no" throws away recoverable ideas.

The first experiment is free, including the market report. See also how to validate a business idea for the steps that come before this one.

theLab builds the landing page, finds real people who match your audience, runs the outreach, and returns a verdict from real signups. First experiment free, no card.