Mention rate

What the mention rate measures, why each prompt is asked several times, and how to read visible, cited and failed runs.

The mention rate is the share of AI answers that name your brand or cite your site for a prompt. AI answers change on almost every run, so AgenticSEO asks each prompt several times per engine and reports a percentage instead of a single yes or no. A prompt counts as visible when the mention rate is 50% or more.

Why one run is not enough

Ask ChatGPT the same question twice and you often get two different lists of brands. Research by SparkToro and Gumshoe found that the same prompt gave the same brand list in fewer than 1 in 100 runs. A single check is noise. A rate over several runs shows a real pattern.

Read more on our blog: Why AI answers change.

How a round works

One round is one prompt asked in one engine Runs per prompt times (default 5, from 1 to 10). For each answer, the agent records:

  • Named: the answer mentions your site name or domain.
  • Cited: the answer links to a page on your domain.
  • Competitors: which of your competitors the answer names or cites.
  • Cited domains: every other domain the answer links to.

A run that is cited also counts as named.

Term Formula
Mention rate runs that named or cited you / runs that returned an answer
Visible mention rate of 50% or more
Cited runs that link to your site

Runs that fail (for example a rate limit) are left out of the rate. If every run fails, the cell shows error. Hover it to read the message.

Note: The agent matches your Site name (Settings > General) and your domain. Set the site name to your brand as people write it, for example "Acme Payroll", not "Acme Payroll Website".

Read the numbers in Insights

The AI assistant prompts card shows one box per engine:

  • The big number is the percent of that engine's runs that name you.
  • Below it: named runs of total runs, how many cite a page, and how many prompts are visible.

Each prompt row shows a badge per engine:

Badge color Meaning
Green Visible (50% or more) and cited at least once
Blue Visible, but never cited
Amber Named in some runs, below 50%
Gray Never named

Next to the badge you see, for example, 3/5 · 1 cited: named in 3 of 5 runs, cited in 1.

Cost of each check

Each check is a real API call with web search on. The cost per week is about:

prompts x runs per prompt x engines = AI calls a week

With 20 prompts, 5 runs and both engines, that is 200 calls a week. The settings page shows the current number per engine under Runs per prompt. These calls are included in your plan. See AI usage.

Tip: Use small models for the checks. The defaults are gpt-5-mini and gemini-2.5-flash. Each check also pays for a web search.

How to raise your mention rate

AgenticSEO does not promise citations. These steps make them more likely:

  1. Answer the prompt early on the page. Gap tickets tell the writer to answer directly near the top, in plain words, with specific facts.
  2. Add facts and sources. The GEO-bench paper (KDD 2024) found that adding statistics, quotes and sources raised visibility up to 40% in tests, while keyword stuffing lowered it.
  3. Keep pages fresh. Ahrefs found that ChatGPT cites newer pages than Google organic results.
  4. Rank in classic search. Seer Interactive found 87% of SearchGPT citations matched Bing top results.
  5. Look at who gets cited instead. The "Other sites the AI cites most" list shows review sites, forums and blogs to target. See Competitors.

Compare over time

The table shows the latest round per prompt and engine from the last 60 days. Click Check now on a prompt after you publish or update its page to see if the rate moves. Give it time: AI engines need to find and read the new page.

Warning: A small number of runs swings a lot. With 3 runs, one extra mention moves the rate by 33 points. Use 5 or more runs before you draw conclusions.

Did this page help? Tell us at support@agenticseo.com.