Worked examples: what a test looks like, start to finish

Every testing tool promises insight. The useful question is what one run actually gives you, including when the answer is unflattering or uncertain. These examples show the whole thing, so you can judge before you run your own.

• A landing page with no traffic yet: where trust and clarity break.
• A sign-up flow: the step that confused people, even when everyone finished.
• Two headlines: when a call is clear, and when it is too close to call.
• What each kind of test cannot tell you.

Input: the Tallyloop homepage, a made-up invoicing tool for agencies. Audience: 25 simulated owners and finance leads at US marketing agencies and service businesses, matched from the page.

What came back: only 9 of 25 stayed to read, 12 skimmed and 4 left. The top issue, raised by 16 of 25, was that no pricing was visible, so people could not tell whether it was for a business their size. Next came positioning: 10 read it as a collections tool rather than a way to run their finances, and 10 at non-agency businesses felt it was not for them.

What it cannot tell you: the conversion rate. It shows which objections are common and why, so you know what to fix before you buy traffic.

Open the full report Test your landing page

Input: the Tallyloop sign-up page and the task "Sign up for the free trial". Five simulated people from the same audience each used it in their own real browser.

What came back: all 5 finished, in 5 to 9 steps. Finishing is not the same as smooth: the business URL field rejected valid addresses without saying what format it wanted, and people were unsure whether they could continue past the error. Both showed up as the steps where people slowed down, with their own words about it.

What it cannot tell you: how many real visitors would give up at that field. It tells you the field is a problem worth fixing, and the rerun shows whether the fix worked on the same people.

Test your sign-up flow

Input: two versions of a headline. The same simulated readers react to both, and Mimiq makes a call: a clear pick, a leaning pick, or too close to call.

What came back varies, and that is the point: when the versions differ a lot, the call is often clear; when they are near-identical, too close to call is the honest answer, and you should write a version that differs more. Headlines are where Mimiq has public, pre-registered evidence from historical A/B tests.

What it cannot tell you: your click-through rate. Use the call to pick what to test with real traffic and to drop weak lines early.

Compare two headlines free See the benchmark Play the headline game

Some tests fail honestly. A site behind a login or a bot check stops simulated people at the door; if it stops everyone, the flow fails and nothing is charged. A page that loads slowly or breaks in a real browser shows up as people who could not see it, not as a verdict on your copy.

Every report says the people are simulated, and every call says how sure it is. When it is not sure, it says so.

Trust and data handling