Experiments
Testing Methodology
Every Grow AI Search experiment follows one protocol: a written hypothesis, a fixed prompt set, a fixed schedule, the same engines on every run, a control group where the design allows one, and a published log of every answer. Results that fail to support the hypothesis are published with the same detail as results that do.
What is fixed before an experiment starts?
The hypothesis, the prompt set, the engines, the schedule and the success measure are fixed and published before the first run. Changing any of them mid-test starts a new experiment.
What is logged for every answer?
- Date, engine and prompt ID
- Whether the brand is mentioned, and whether it is cited with a link
- The cited URL and the position of the first mention
- Competitors mentioned or cited
- Sentiment of the mention
The logging sheet is the same one offered free on the AI visibility page.
How are results published?
Each experiment page publishes the raw log as a downloadable file, the summary numbers, and the limits of the test. AI answers vary between runs, so every result states how many runs it rests on.
See the list of experiments.