Test Lab shows your real creative — an image, a carousel, a video or a text — to a panel of simulated people and collects their reactions. It is a way to compare ideas before you spend. It is a simulated reaction, not a sales forecast.
Audiences and panels
Describe who a brand sells to and Reachio builds a panel of 10–50 fictional personas. Their traits are sampled in code with a fixed seed, so the same audience gives the same panel; at least one in five is a skeptic by design. Choose the panel's language and country. A panel is versioned: a test always records which version it used.
Kinds of test
- Single score: one subject, one simulated score.
- Survey: ask open, multiple-choice or A/B questions.
- A/B: two subjects; the same personas see both.
- Experiment: 2–6 subjects across up to 4 panels; a winner is named only when the confidence interval excludes zero.
The cost is estimated before a test runs, and it comes from the plan's AI allowance.
Reading the result
Each persona answers under a strict format. Reachio combines six dimensions — attention, relevance, emotional resonance, persuasion, brand fit and clarity — with weights per platform into a score from 0 to 100 and a band (poor, average, good, strong, exceptional). Every score comes with the personas' written feedback.
Improve and compare with reality
Improve makes new variants aimed at the weakest dimensions and re-tests them against the original with the same panel; Auto loop repeats this for a few rounds. When a tested creative runs as a real ad, the Lab board compares Lab scores with the real click-through rate, so you can see how far to trust them for that brand.