What Is the Answer Score? How Saddle Rank Rates a Page for AI Answers
The answer score rates how easy one page is to quote in an AI answer. See the nine signals and what each is worth.
The answer score rates one page out of 100 for how easy it is to quote in an AI answer. Some people call this an AEO score, for answer engine optimization. It is a fixed list of nine signals. No AI runs to work it out, and every point has a reason you can read.
It is a different number from the readiness score. The readiness score rates your whole site. The answer score rates one page, and the site’s average feeds 35 points of the readiness score.
Where do I see it?
- In the editor. The Saddle Rank sidebar shows the score under AI answer readiness, with one line per signal. It updates when you save.
- For the whole site. Open Saddle → Rank → AI visibility → Answer score. The Answer scores page lists your most recently updated pages, lowest score first, with each page’s weakest signal. Fix in editor opens the page. Score again rescans them.
What are the nine signals?
| Signal | Points | What earns them |
|---|---|---|
| Answer up front | 24 | A direct answer in the opening paragraph, before the first heading. About 40 words or more. |
| Easy to scan | 16 | Short sentences, and a subheading every few paragraphs so each section answers one thing. |
| Numbers and data | 15 | Specific figures in the text. |
| Up to date | 12 | The page changed within the last year, and readers can see when. |
| Cited sources | 10 | Links out to trusted sources for your key claims. |
| Quotes | 10 | Direct quotes from a source or an expert. |
| Question headings | 6 | Section headings phrased as the questions readers ask. |
| Named author | 4 | A visible author bio. |
| Questions answered | 3 | Question-and-answer pairs on the page. |
The nine add up to 100. Most signals earn full, half or no points, and each one comes with a short note that says what to do.
How is “Up to date” scored?
- Changed within a year, and the page shows the date: full points.
- Changed within a year, but readers cannot see when: half.
- Last changed over a year ago: half.
- Not changed in over two years: a quarter.
A new date alone does not count as a refresh. Update the facts first, then the date.
Why are the points split this way?
The weights follow the published evidence, and the strongest evidence gets the most points.
- Where the answer sits matters most. One 2026 study found that 44% of ChatGPT citations come from the first 30% of a page. Bing’s guidelines also ask for key facts near the top.
- Fresh pages are cited more. A 2026 study of 47,000 citations found that 75% of cited pages had been updated within a year.
- Numbers, sources and quotes help, but less than first claimed. An early study measured large gains for each. Later studies found those gains shrink in real search, so these signals carry fewer points.
- Question headings, a named author and FAQ pairs have weak or mixed evidence. They carry the fewest points.
The weights are a guide, not a law. They may change as the evidence does.
What does the score leave out on purpose?
- Word count. Google says there is no ideal length.
- Keyword stuffing. It has no effect or does harm.
What can the score not see?
A page score only reads the page. It cannot see two things that matter a lot:
- How well the page already ranks in search.
- How often other sites mention your brand.
So a high score does not mean an AI tool will cite the page. It means the page is easy to quote if it is found.
How do I raise a page’s score?
- Answer the main question in the first paragraph.
- Use headings that ask the questions readers ask.
- Keep one idea per paragraph, so each passage stands on its own.
- Add real figures, and link each one to its source.
- Show an “Updated” date, and keep the facts current.
Write for people first. A page written only to fill the list reads worse.
Common questions
Does the answer score affect my Google ranking?
No. It is Saddle Rank’s own measure. No search engine or AI tool reads it.
Why is a good page scoring low?
The score looks for specific things: an early answer, figures, sources, quotes, question headings. A well-written essay with none of them scores low without being bad writing. Read the notes under each signal to see what is missing.
Can my AI agent read the score?
Yes. An agent connected through Saddle can read a page’s score and the reasons behind it.
Sources
- GEO: Generative Engine Optimization (Aggarwal et al., KDD 2024): arxiv.org/abs/2311.09735
- Ahrefs, search rankings and AI citations: ahrefs.com/blog/search-rankings-ai-citations