How to Measure AI Search Visibility
NAFETS TEAM
2026-03-16
If you only watch Google rankings, you will misread AI search. Citation share on a fixed prompt set is how we measure AI SEO progress at Nafets.
Build a prompt set (20–30)
Cover the journey:
- Category: “best AI SEO agency for SaaS”
- Problem: “how to get cited in ChatGPT”
- Comparison: “Nafets vs [competitor]” (when fair)
- Product: “best [your category] for [job to be done]”
Freeze the wording for 90 days so month-over-month is comparable.
Score each run
For every prompt × surface (ChatGPT, Perplexity, Gemini, AI Overviews):
| Score | Meaning |
|---|---|
| 0 | Absent |
| 1 | Mentioned without recommendation |
| 2 | Cited / linked |
| 3 | Recommended as a top choice |
Note the competitors who win when you score 0–1.
Cadence
- Weekly (optional): spot-check top 5 money prompts
- Monthly: full set, same operators, same models where possible
- Quarterly: refresh 10–20% of prompts as language shifts—retire, don’t silently edit
Keep a simple spreadsheet or database. Fancy tooling helps later; consistency matters first.
Tie metrics to actions
- Low mentions → entity and off-site authority gaps
- Mentions but no citations → weak extractable owned pages
- Citations but no recommendations → comparison and proof gaps
Next steps
Use our internal-style template in the repo (docs/ai-seo-prompt-set.md) as a starting sheet, then operationalize it under AI SEO. Classic SEO metrics still live in Search Console—run both.
Inquire if you want Nafets to own the monthly measurement loop.
