Measure · Google and AI, one row per question

Search layer

Why does a page that ranks #2 on Google still get skipped by ChatGPT? The Search layer puts the two results side by side, so you can see it.

What is the MentionX Search layer?

The Search layer turns every buyer prompt in your scan into a search twin, the keyword a buyer would type into Google, fetches Google's organic top ten for it, and places your rank next to the AI answers for the same question. Code sorts each question into one of four states, win both, win in AI only, win in Google only or lose both, and attaches one fix.

See my Google-vs-AI rowsSee the methodology
All plans
How it works

How Search layer works: How the Search layer works

1Each prompt gets a search twin

"Best CRM for a 200-person sales team?" becomes "best crm for mid-size sales team". Twins are checked against the domains AI cites for the same question; a weak twin is flagged, never hidden, and you can edit it.

2Google's top ten is fetched and stored

Through a licensed search-results API, no personal accounts. Your rank, your page and whether Google showed an AI Overview are stored as evidence beside the raw AI answers.

3Both results are compared in code

No model decides the outcome. Rank in the top ten plus an AI recommendation is "win both"; AI yes and Google no is "win in AI, lose in Google"; and so on for all four states.

4The quoted page is diagnosed against yours

When AI quotes a rival's page instead of yours, MentionX fetches both and lists the difference: a direct answer in the first screen, a comparison table, proof, an FAQ.

5The fix becomes a brief

Make quotable, improve rank, create page or defend. Each opens a brief in the GEO Content Agent, and next week's scan measures the same question again.

Read it with

Metrics related to Search layer: Three numbers that tell the full story.

Google top-10 rate
questions where your page is in the organic top 10 ÷ questions measured

Your search footprint on the questions buyers actually ask, not on a keyword list.

Recommendation Rate
AI answers that recommend you ÷ all answers

The AI side of the same row. Read both together: a high Google rate with a low Recommendation Rate is the classic "ranks but silent" pattern.

Citations

The exact page each engine quoted. When it is a rival's page for a question you rank for, the diagnosis shows what their page has that yours lacks.

Rank

Rank tracker vs the Search layer: A rank is not a measurement of the buyer's question.

Rank tracker

"crm mid-size" · position 2 · 1,900 searches a month. Nothing about what ChatGPT, Gemini or Perplexity answer when the same buyer asks.

MentionX Search layer

Best CRM for a 200-person sales team? · Google #2 · AI: not recommended, Kestrel's comparison page quoted · Fix: make the page quotable

Same question. The second line tells you what to change on Monday.

Raise it

How to improve your Search layer: From the number to the fix.

MeasureRecommendation Rate

Recommendation Rate is the share of AI answers to real buyer prompts in which an engine recommends your brand as an opti…

Explore →UnderstandCitation & Source Intelligence

AI citations are the web pages an engine links to or relies on when it answers. MentionX stores every cited URL, so you …

Explore →FixContent Gap Analysis

AI content gap analysis finds the themes, sources and claims engines use when they recommend competitors instead of you,…

Explore →FixGEO Content Agent

A GEO content agent drafts content designed to be cited and recommended by AI engines. MentionX's GEO Content Agent writ…

Explore →
Search layer FAQ

Search layer FAQ: Questions about the metric.

How it is counted, what a good number is, and how to move it.

No. It tracks the buyer questions in your scan, not keyword lists, and its purpose is the comparison with AI answers. Keep your rank tracker for keyword portfolios; use the Search layer to see which questions you rank for and still lose in AI.

Google organic results, top ten per question, fetched through a licensed search-results API for the market your scan runs in. MentionX also records whether Google showed an AI Overview for the query.

The short query MentionX derives from a conversational buyer prompt, what the same buyer would type into Google. Every rank on the Search tab is for the twin, not the full prompt. You can edit any twin.

Not yet. Google results come from a licensed search-results API, so there is nothing to connect. A Search Console import is on the roadmap.

Win both: you rank and AI recommends you. Win in AI, lose in Google: AI recommends you but your page is not in the top ten. Win in Google, lose in AI: you rank but AI skips you, usually because the page is not quotable. Lose both: no page of yours answers the question yet.

Yes. It runs on every plan as part of the weekly scan, on every buyer prompt, with no separate search credits.

Start free

What is your Search layer?

Enter your URL and market. Your first scan shows it on every engine in your plan.

All features