Rank AI: health score, evidence and what to fix first
How the ranking health score is built, how to read recommendations and their evidence, and when to refresh.
- Lesson
- 7/9
- Reading time
- 6 min
- Sections
- 4
- Through the course
- 78%
One score from several modules
Rank AI uses the data your other modules already collect for a website: positions, SEO audits, uptime incidents, AI Overview citations, local results, analytics traffic and lost backlinks. The health score runs from 0 to 100, and sources without data are left out rather than counted as zero.
Reading a recommendation
- Each recommendation has a severity and shows its evidence, such as a position change, an audit issue or an incident.
- A position drop with an uptime incident from 3 days before to 1 day after it is marked, so you can check the outage first.
- Recommendations come from fixed rules, so the same data always gives the same list.
Check your understanding
A keyword slipped from 4 to 8 and an uptime incident is shown next to it. What should you check first?
Page reviews
- Pick one of your pages and the search term it should rank for. Rank AI compares it with the top-ranking pages your runs have recorded for that term.
- Lime arrows on the copy of your page mark what to change. Each suggestion quotes your text, says why, shows the evidence and gives a rewrite to copy.
- Start with high-priority suggestions and missing topics several competitors cover, then review the page again after editing to compare the score.
Opportunities and the weekly summary
Opportunity keywords sit in positions 4 to 20, where small improvements move the most. Every week a report is saved with a short summary; AI writes it from the report's numbers only, and a summary that can't be checked against the data is replaced with a plain one.
Lesson 7 of 9
Rank AI: health score, evidence and what to fix first