Every figure shows its sample
A percentage is printed with the number of answers it was measured over. 11% from 9
answers and 11% from 900 are different claims, and you can see which one you have.
A change needs evidence
rekkal only draws an arrow when the two periods are far enough apart that sampling
noise cannot explain the gap. Otherwise it says No measurable change.
Absence is shown as absence
A week nothing ran is a hole in the chart, not a zero. A missing comparison says which
half is missing.
Why that matters more than it sounds
AI answers are not stable. The same question, asked twice, frequently returns a different list of brands. Any tool that reports a single number from a single sample is reporting mostly noise — and if it draws that number as a trend line, it is inventing a story. rekkal is built the other way round: repeat the same prompts on a fixed schedule, count the answers, and refuse to claim a movement the count cannot support. The reasoning, the independent research behind it, and the arithmetic are on the Methodology page.Start here
Set up your brand
Type one domain. rekkal derives your brand profile, topics and prompts; you review and
approve.
What rekkal measures
Visibility, share of voice and average position — what each one counts, and what it is
divided by.
Reading a change
Why a number moving is not always a change, and what each verdict on the card means.
Plans and limits
Prompts, engines, tracked competitors, and what Pro adds.
rekkal is pre-launch. Pages here describe what the product does today; anything not yet
built is marked Not available yet rather than described as if it were.