Do this next
Every reading on your domain page is a signal. I fold them into one ordered to-do list, with the item that moves your AI presence the most at the top. There's no guesswork here: same signals in, same list out. No model decides the order — a fixed set of rules does.
One rule per signal. Each turns a reading from another card into a plain recommendation, carries the evidence that proves it (the "why"), and links back to the card where you can see it for yourself.
| Signal | Recommendation | Priority | Evidence |
|---|---|---|---|
| Blocked citation crawlers | Unblock the named bots in your robots.txt | high | Access — you're invisible to those engines |
| JavaScript-only pages | Server-render those page(s) | med | Access — their content can't be read |
| Worst AEO dimension | Improve that dimension on your pages | high if it scores below 50, else med | Readiness — your biggest AEO drag |
| Top not-cited questions (up to 3) | Win each question you're not cited for | high or med — the question's own opportunity | Citation — a rival is cited, or no one is |
The AEO-dimension rule is the only one with a threshold: a score under 50 makes the fix urgent (high); at 50 or above it's a med.
The list is sorted by priority alone: every high item first, then every med item. When two items share a priority, they keep signal order — access first, then readiness, then citation. Nothing else reorders the list.
High before med; within a tier, access then readiness then citation.
I rank, I don't promise. On this to-do list the impact stays a priority, not a number — I never bolt a "+N points" estimate onto an action here. The one place I do quantify a fix is the Readiness card, where each win shows the points it would recover, grounded in the score math. The ranking itself is deterministic: an honest "this matters most" beats a made-up number.
Every row points back to the card that holds its evidence, so you can check the reading before you act: