It has never been easier to buy a signal. A sharp headline, a clear landing page, and a small distribution budget can produce visitors, clicks, and even sign-ups before a product really exists. That is useful. It is also dangerous, because the dashboard can look like evidence long before the experiment has tested the decision we care about.
I like cheap tests. They protect builders from spending months on an idea nobody notices. The mistake is not running them. The mistake is asking an attention experiment to answer a product question.
Every signal sits in a chain
An impression can test whether a message reaches someone. A click can test whether the promise creates curiosity. A sign-up can test whether that curiosity survives one more step. None of them proves that the person can use the product, receive the promised value, or care enough to return.
Teams often collapse this chain because the early numbers arrive quickly. A low cost per click feels precise. Conversion percentages invite comparison. Yet precision is not the same as relevance. If the product decision is whether to invest in a workflow, the useful evidence must get closer to the workflow than the advertisement.
A cheap signal can tell you where to look. It cannot tell you what you will find there.
Match the evidence to the decision
The right question is not whether a metric is good. It is whether the metric can change the decision in front of the team. If the team is choosing positioning, message response matters. If it is choosing onboarding, first-use completion matters. If it is choosing whether the product deserves continued investment, repeated value matters much more than the first click.
This also explains why one-variable tests are so valuable. Change the audience, promise, page, price, and call to action at once, and a winning result teaches almost nothing. A clean experiment can feel slower because it explores fewer ideas per round, but it produces knowledge the next round can actually use.
Each rung answers a different question. Skipping upward is fine when the experiment is explicitly exploratory. The trouble begins when a team stands on the attention rung and tells itself it has measured retention, willingness to pay, or durable demand.
A failed test can still be useful
Cheap experiments become much more honest when the failure interpretation is written before launch. If nobody clicks, will the team reject the problem, the message, the audience, or the channel? Those are different conclusions. Without that agreement, results become a storytelling exercise and every number can be explained after the fact.
The same discipline applies to success. A strong click result should earn the next test, not the whole roadmap. It says the promise deserves a closer look. The next step might be a manual service, a narrow prototype, or a workflow people complete with real inputs. Evidence should increase in cost as the decision becomes harder to reverse.
The purpose of a cheap click is not to prove the product. It is to avoid paying full price for the next piece of uncertainty. That is already valuable. We only lose the value when we ask the click to carry a conclusion it was never designed to support.