AI Visibility Audit Template (Sheet + Report)

This AI visibility audit template is a spreadsheet and one-page report structure for checking how often AI engines mention, cite and recommend your brand compared with competitors. You log prompts, run them in each engine, record what appears and which sources are cited, then summarise the findings. It works manually or with a MentionX scan export, and fits a first audit of 20–50 prompts.

What does the template do?

It gives you a consistent way to answer four questions:

1. Do AI engines know about us? (mention rate)
2. Do they use our content as a source? (citation rate)
3. Do they recommend us? (recommendation rate)
4. Who wins instead, and why? (competitor share and cited sources)

Consistency matters more than tooling. If you record results the same way every time, you can compare audits month to month.

Who is this template for?

It suits marketers running their first GEO check, agencies preparing a pitch or onboarding a client and in-house teams that want a repeatable monthly baseline before buying tools. Because every field is defined, two people can fill it in and get comparable results, which matters when an audit is shared across a team or handed over to a client.

How long does an audit take?

As a rough estimate, a 20-prompt audit across three engines with one run each produces 60 answers and takes three to four hours by hand, including recording sources. Doubling to two runs roughly doubles the recording time. A scan export removes most of the manual logging, leaving your time for analysis and the report, which is where the value is. Block the time in one sitting if you can, so every answer comes from the same day.

What's in each tab?

What do the fields mean?

Prompts tab

Results tab

Sources tab

How do you fill it in?

• [ ] Enter your brand, three real competitors and the engines you'll check in Setup.
• [ ] Write 20–50 prompts across all four stages in Prompts.
• [ ] Run each prompt in each engine. Use a clean session, without signing in where possible, and note your location.
• [ ] Record every answer in Results; repeat runs if you can.
• [ ] Copy every cited URL into Sources.
• [ ] Check the Summary tab and write the Report.

If you use MentionX, paste the scan export into the Results tab. The columns are designed to match.

What does a filled example look like?

Illustrative example only; the brand and results are invented.

Summary for this fictional audit: mentioned in 2 of 4 answers, cited in 1, recommended in 0. With only four rows, that's a format example, not a finding.

How are the summary rates calculated?

Report rates by engine as well as overall. One engine often behaves very differently from another. With small samples, add a range: the NIST guidance on the Wilson interval explains a sound method for proportions.

How do you turn it into a report?

Use the one-page outline in the Report tab:

1. Headline: one sentence on where you stand.
2. By engine: a small table of the three rates.
3. Competitors: who wins and on which prompt types.
4. Sources: the five most-cited domains and whether they mention you.
5. Accuracy issues: wrong prices, outdated features, confusion with other brands.
6. Three actions, each tied to a finding.
7. Limits: sample size, date and location.

What mistakes should you avoid?

• Leading prompts. Don't ask "Is [brand] good?" in category research.
• Mixing methods. Keep the engines, settings and location consistent between audits.
• Single runs only. Answers vary; even two runs show you how much.
• Skipping sources. The Sources tab is where most actions come from.
• Overclaiming. A first audit is a baseline, not proof of a trend.

How often should you re-run the audit?

Monthly works for most teams; quarterly is fine for slower-moving categories. Keep the same prompt set as your core and add new prompts in a separate section, so your trend line stays comparable.

Where next?

Learn to design better prompts in GEO 101, go deeper with the GEO audit template and the citation gap analysis template, or use this template for the Break into GEO capstone.