Founders defer tests because they 'cost too much.' The math almost always says the opposite — here's the back-of-envelope every PM should know.

Every roadmap has a few assumptions doing very heavy lifting. 'Users want this.' 'Pricing is fine.' 'The onboarding is the problem.' Each one shapes months of work, and almost none of them get tested before the work starts, because testing 'feels expensive.'

Run the numbers. A landing-page test with paid traffic to validate demand: ~$500 and 3 days. A 5-user concierge test of a feature before you build it: ~$0 and one week. A pricing test on the next 200 signups: free and instant. Now compare to the cost of being wrong: an engineer-quarter is ~$60k loaded. A misaligned positioning rewrite is ~$30k in agency fees plus a month of stalled growth.

The ratio is almost never close. The test costs 1–5% of the cost of being wrong, and it resolves the question in days instead of quarters. The reason teams skip it isn't economic — it's emotional. Tests can return inconvenient answers. Building lets you stay in motion and feel productive while the question stays unresolved.

The discipline to install: every roadmap item over two engineer-weeks gets a one-line 'cheapest test that could kill this' attached to it. If the test is cheaper than the build (it almost always is), you run the test first. If you can't think of a test, that's a signal the assumption isn't actually testable — which is a much bigger problem than the build estimate.