Reference table
The experiment cost ladder: run the cheapest test first

A table of the experiments a product team can use to test a hypothesis, ordered from cheapest to most expensive: customer interviews, surveys, pretotypes, prototypes, MVPs and A/B tests, and finally a full launch. Start with the cheapest test that could prove the idea wrong and only move up when it can't settle the question. Koji covers the first two rungs, with AI-moderated interviews and structured questions in the same study.
Where this appears
What's in this reference table
| Experiment | Relative cost | What it tells you |
|---|---|---|
| Customer interviews | $ | Whether the problem and the proposed solution hold up |
| Surveys and structured studies | $ | How many customers share the problem |
| Pretotypes | $ | Interest, from fake doors, landing pages or Wizard of Oz tests |
| Prototypes | $$ | How people use a clickable mockup in a usability test |
| MVPs and A/B tests | $$$ | How a minimal real build does behind a feature flag |
| Full feature launch | $$$$ | Last resort, once the evidence is clear |
Use this image
Free to use in your docs, decks and posts under CC BY 4.0, as long as you credit Koji with a link to this page. To embed it, paste this into your page:
html
<a href="https://www.koji.so/visuals/experiment-cost-ladder-cheapest-test-first"><img src="https://www.koji.so/visuals/experiment-cost-ladder-cheapest-test-first/experiment-cost-ladder-cheapest-test-first.png" alt="Table ranking six product experiments by relative cost, from customer interviews and surveys up to prototypes, MVPs, A/B tests and a full launch." width="1200" height="877" /></a>
<p>Source: <a href="https://www.koji.so/visuals/experiment-cost-ladder-cheapest-test-first">The experiment cost ladder: run the cheapest test first</a>, Koji</p>




