What it does
Builds a period-over-period comparison (choice of 7/30/90-day windows) of your real recorded metrics against the immediately preceding period of equal length. A change is only marked as significant once it clears normal measurement noise -- small fluctuations are shown as such, never presented as real movement.
Shows an AI-visibility-score trend chart (needs at least two weekly scoring runs to render), a period-comparison table (prior value, current value, delta -- rows explicitly say 'Not enough data yet' when the prior period has nothing comparable), a by-provider breakdown (runs/mention-rate/recommendation-rate per AI engine), a by-language breakdown (English vs. Hindi/Hinglish), tracked-competitor movement (newly tracked / no longer observed, with last-seen dates), and recorded AI-crawler access changes over the window.
Filterable by period, provider, and language.
How to use it
- 1Pick a time window (7/30/90 days).
- 2Optionally filter to one AI provider or one language group.
- 3Read the trend chart and period-comparison table -- rows marked 'Not enough data yet' mean there's nothing comparable in the prior period.
- 4Use an evidence link, or 'See before/after proof for a specific action', to jump to Proof for a specific change.
Known limits
- The exact statistical method behind what counts as a 'significant' change wasn't independently verified at the formula level -- only that the gate exists and controls what's shown as real movement vs. noise.