Every answer is right or it isn't. Right means consistent with your verified facts. Not-right splits into wrong (contradicts them), outdated (behind a real change), and missing (no usable answer). A verdict only changes after two consecutive scans agree, so one flaky response never moves your numbers.
Each flag's exposure = the question's estimated monthly volume × the model's share of AI usage. Errors on big questions in big models rise to the top because more buyers hear them. The arithmetic sits on every flag and is always labeled an estimate.
No composite score, no invented weights, no 78-out-of-100 to take on faith. These numbers move only when an answer actually changes, every change carries a date and the answer that caused it, and question-set changes are versioned in your history.
45 questions: 30 brand facts, 15 category and comparison. Daily delta covers flagged questions plus the top 14 by volume.
Grow builds the presence you don't have yet: new citations, content models recommend, visibility where AI actually looks. Scoped to your niche.
Book a Grow call →Every fix you approve is implemented by our team, joins the daily scan set, and shows as verified after three consecutive clean reads. Prefer to handle one in-house? Mark it done and we verify it the same way.
Models take time to re-read corrected sources, so a fix can be real days before the answers move. This log is the record of the work itself, independent of the scan cycle.
Demo data. All numbers are sample values; on a live dashboard every count links to the answers behind it.