Scores & movement
I report your standing as numbers you can act on. Two of them I measure natively, on a 0–100 scale; the third is their blend. Every score carries a colour that says what it means, a benchmark that says how far to the next tier, and — once I have read you twice — a delta that says which way you're heading.
Your scoreboard is two native scores and one blend. Readiness is on-page: how citable your pages are when an assistant reads them — the plain mean of your individual page scores, where every page counts equally, because that is simpler and more honest than a weighting I'd have to defend or a single worst page speaking for the whole domain. Citation is off-site: the share of answers that name you when engines are actually asked. AI presence is the plain mean of the two — a single headline for where you stand.
Illustrative figures. AI presence = mean(Readiness, Citation) = mean(64, 42) = 53. As more stages of the stack go live the blend widens; today it is these two.
Any 0–100 score also reads as a colour. The same three tokens apply everywhere a score appears in the product, so a glance tells you whether a number is good, worth watching, or needs work.
A lone number can still feel ambiguous, so I pair it with a benchmark: how far to the next band. A score of 80 or more simply reads strong. A score in the watch band reads N to strong — the points up to 80. A score below 50 reads N to fair — the points up to the watch band. So the 64 above reads "16 to strong", and the 42 reads "8 to fair".
Each scoring pass appends a snapshot — that pass's Readiness and Citation reading; AI presence is derived from the pair, not stored separately. The scoreboard then shows the delta: the point change since the previous pass.
A delta appears only once two snapshots exist — there is no backfill. On the very first pass there is no earlier measurement to subtract from, so I show the score with no movement rather than invent a baseline. The arrow you see later is a real change between two real readings.
This page is about how the scores read and move. How they are computed lives with each metric, and the snapshots that record movement are written by the background passes: