Skip to content
Chalk Labs / AI SEO
How-to guide

How to Set Up an AI Visibility Tracking Spreadsheet

By Rahil Jain, founder of Chalk LabsUpdated How we research

The short answer

To track AI search visibility manually, build a spreadsheet with four tabs: Prompts, Runs, Sources and Summary. Run the same 30 to 40 prompts in each AI tool once a month, log whether you were named, cited and accurate, and let formulas turn the Runs tab into three trend numbers: mention rate, citation rate and accuracy rate.

  • Four tabs are enough: Prompts, Runs, Sources and Summary. Formulas in Summary do the maths.
  • Run 30 to 40 fixed prompts monthly. Prompt choice matters more than run frequency.
  • Track three rates: mention, citation and accuracy. Keep notes for tone and wrong facts.
  • Manual tracking suits small teams. Move to a paid tool when you need daily runs or hundreds of prompts.

What is an AI visibility tracking spreadsheet?

An AI visibility tracking spreadsheet is a table where you log what AI assistants say about your brand for a fixed set of prompts, month after month. A prompt is the question you type into an assistant. Visibility means how often you are named or cited in the answer.

It does the job of a paid tracker at small scale. Ahrefs puts the core method simply: track a sample of prompts and see who gets recommended 1. A sheet lets you do that for free, and it forces you to read the answers instead of looking only at a dashboard score.

If you have not yet looked at your starting position, run the one-off brand audit in AI search first. Its output becomes month zero in this sheet.

Do it step by step

0 of 7 done0%

  1. Make a new spreadsheet and name the tabs Prompts, Runs, Sources and Summary. Freeze row 1 on each. Paste the column headers from the templates below into row 1 of Prompts and Runs.

  2. Add an id (P01, P02 and so on), a group (brand, category, comparison, problem), the prompt text and a priority of high, medium or low. Aim for a spread across all four groups.

    Tip: Lock the Prompts tab after month one. Add new prompts only at the bottom, with a start month.

  3. Pick three to five AI tools, such as ChatGPT, Perplexity, Gemini and Google AI Mode. Pick one day each month, for example the first Tuesday, and put it in your calendar.

    Tip: Run all tools on the same day, so a model update does not split your month in two.

  4. Start a new chat for every prompt. Paste the month, tool and prompt id into the Runs tab, then fill named, position, cited, accurate and competitors. Add a note for wrong facts and odd tone. Save the answer text or a screenshot in a linked folder.

  5. For each run, copy the domains that were cited. Add one row per domain per month with a count. Tag each domain as own, competitor, directory, review, news or forum.

  6. In Summary, list each month in column A as text, such as 2026-11. Paste the three formulas from the template below in B, C and D. Insert a line chart of the three rates.

  7. Each quarter, sort Runs by accurate = no and by prompts where a competitor is named and you are not. Turn the top three patterns into tasks. Record any site changes you made in a Changes tab with dates, so you can link cause to effect later.

    Tip: Keep a short change log. Without it you will not know what moved the numbers.

Your ticks are saved in this browser only.

Copy-ready templates

Header rows for the Prompts and Runs tabs

text
PROMPTS (row 1)
id,group,prompt,priority,added_month

RUNS (row 1)
month,tool,prompt_id,named,position,cited,accurate,competitors,notes

VALUES
month: text such as 2026-11
named: yes or no
position: first, top3, lower or none
cited: yes or no (your own domain linked as a source)
accurate: yes, no or na (na when not named)

Use data validation drop-downs on named, position, cited and accurate so entries stay clean.

Summary formulas (Google Sheets or Excel)

text
Mention rate (cell B2, month in A2):
=COUNTIFS(Runs!$A:$A,$A2,Runs!$D:$D,"yes")/COUNTIFS(Runs!$A:$A,$A2)

Citation rate (cell C2):
=COUNTIFS(Runs!$A:$A,$A2,Runs!$F:$F,"yes")/COUNTIFS(Runs!$A:$A,$A2)

Accuracy rate (cell D2):
=IFERROR(COUNTIFS(Runs!$A:$A,$A2,Runs!$G:$G,"yes")/COUNTIFS(Runs!$A:$A,$A2,Runs!$D:$D,"yes"),"")

These assume the Runs column order above: A month, D named, F cited, G accurate. Format B to D as percentages.

Monthly run checklist

markdown
- [ ] Same day, same tools as last month
- [ ] New chat for every prompt
- [ ] Logged out or private window, same country
- [ ] One row per run, answers saved
- [ ] Sources tab updated
- [ ] Summary formulas copied down one row
- [ ] Changes tab updated with anything we published
- [ ] Three patterns written down

Print it or paste it into your task tool.

Why use a fixed prompt list, and how often should you run it?

Use a fixed list because the list is your measuring stick. If you change the questions, you cannot tell whether a change in results came from your site or from your wording.

Run it monthly for most small teams. Profound's June 2026 study concluded that which prompts you track matters more than how often you run them 2; the figures are in the brand audit guide. That study tested one versus ten runs per day, not monthly checks, so monthly is our practical suggestion for slow-changing small sites, not a tested interval. Profound sells daily tracking, so weigh its conclusions accordingly.

The same study names three reasons results move: random text generation, sensitivity to rewording, and platform drift after model updates 2. A sheet cannot remove that noise. It can show you whether the trend over several months is bigger than the noise.

What should each tab contain?

Each tab has one job. Prompts holds the question list. Runs holds one row per prompt, per tool, per month. Sources counts the domains that get cited. Summary holds the formulas and the chart.

Keep raw data in Runs and never edit it after the month closes. Everything else is calculated from it, so mistakes are easy to trace.

The four tabs
TabOne row isKey columnsUsed for
PromptsOne promptid, group, prompt text, priorityThe fixed list
RunsOne prompt in one tool in one monthmonth, tool, prompt id, named, position, cited, accurate, competitors, notesRaw results
SourcesOne cited domain in one monthmonth, domain, count, typeSeeing who owns the sources
SummaryOne monthmention rate, citation rate, accuracy rate, top competitorTrend and chart

Which numbers should you track?

Track three rates: mention rate, citation rate and accuracy rate. Mention rate is the share of runs where your brand is named. Citation rate is the share where a page on your own domain is linked as a source. Accuracy rate is the share of runs that name you where every fact is right.

Keep these separate. Seer Interactive argues that a single visibility figure counts a positive and a negative mention the same, and suggests measuring whether answers are accurate as a separate tier 3. Accuracy catches a problem that a mention count would hide.

Add a position note when you appear in a list: first, top three, or lower. Do not invent a weighted score. With 30 to 40 prompts the data is too thin for fine scoring, and the simple rates are easier to explain to a boss.

What does a month of results look like?

This example is invented to show the arithmetic, not to suggest a benchmark. Say a small accounting firm tracks 30 prompts in 4 tools. That gives 120 runs per month. In month one the firm is named in 18 runs, cited in 9, and gets every fact right in 12 of the 18 named runs.

Mention rate is 18 divided by 120, which is 15%. Citation rate is 9 divided by 120, which is 7.5%. Accuracy rate is 12 divided by 18, which is 67%. The accuracy figure says a third of mentions contain an error, so fixing facts comes before chasing more mentions.

Illustrative Summary tab (invented numbers; your first three months may not rise)
  • Month Runs Named Cited Accurate Mention Citation Accuracy
  • 2026-11 120 18 9 12 15.0% 7.5% 66.7%
  • 2026-12 120 22 12 18 18.3% 10.0% 81.8%
  • 2027-01 120 21 11 19 17.5% 9.2% 90.5%

What are the limits of manual tracking?

Manual tracking has three limits. It is small, with roughly 100 to 200 runs a month. It is slow, since in our experience each run takes roughly a minute or two. And it does not show real traffic, only what assistants say when you ask.

Pair it with data from your own tools. Google Analytics 4 has a default AI Assistant channel for visits from assistants such as ChatGPT, Gemini and Copilot, and it counts Google AI Overviews and AI Mode under Organic Search instead 4. Bing Webmaster Tools also has an AI Performance report, but Microsoft says its grounding-query data is a sample of overall citation activity 5. See how to track AI traffic in GA4.

Ahrefs warns that many AI platforms strip referrer data, so some AI visits land in direct traffic, and it suggests adding AI tools to a how-did-you-hear-about-us question 1. Add that question to your forms. It catches buyers who read an answer and then searched for you by name.

When should you move to a paid tool?

Move to a paid tracker when you need more than about 50 prompts, daily runs, or several markets. The sheet still helps you decide which prompts matter, which is the hard part. A tool then runs them at scale.

Until then, the sheet is enough. The prompt list builder can draft your first 30 to 50 prompts, and the guide to building a prompt list for tracking explains how to group them.

Frequently asked questions

How do I track AI search visibility manually?

Keep a fixed list of 30 to 40 prompts, run them in each AI tool on the same day every month, and log whether you were named, cited and accurate. Use simple formulas to turn those rows into three monthly rates, and chart them over time.

How many prompts do I need?

Thirty to forty is a workable minimum for a small business. In our judgement, fewer than that and one odd answer swings the rate. Profound's June 2026 study found that the choice of prompts moved results more than the number of daily runs 2.

Should I run each prompt several times?

Not for a monthly sheet. Profound found a once-a-day reading was within about two percentage points of a ten-runs-a-day reading on visibility 2. You gain more by adding different prompts than by repeating the same one.

Can I see AI traffic in Google Analytics?

Partly. GA4 has an AI Assistant channel for visits from assistants like ChatGPT and Gemini, but it files Google AI Overviews and AI Mode under Organic Search 4. Some AI visits also arrive as direct traffic, so treat the numbers as a floor.

What is a good mention rate?

No reliable public benchmark exists for small businesses, so we will not give one. Compare yourself with your own first month and with the competitors you named. A rising trend over three months matters more than any single figure.

About the author

Rahil Jain

Rahil Jain is an AI SEO (answer engine and generative engine optimisation) consultant and co-founder of Chalk Labs.

LinkedIn · Method and credentials

Sources

  1. How to Measure AI Search Visibility When Attribution Falls Short · AhrefsOpened October 20, 2026. Supports: Track a sample of prompts; many AI platforms strip referrer data so visits appear as direct; add AI tools to a how-did-you-hear question.
  2. Is once a day enough? · ProfoundOpened October 20, 2026. Supports: 753 prompts across 7 platforms, 1 to 14 June 2026; visibility 78.7% once daily vs 80.4% at ten runs; three sources of variation; prompt choice matters more than run frequency.
  3. AI Visibility Is Lying to You: There's No One-Size-Fits-All AI Search KPI · Seer InteractiveOpened October 20, 2026. Supports: A visibility metric counts positive and negative mentions the same; proposes answer accuracy as a separate measure.
  4. [GA4] Default channel group · Google Analytics HelpOpened October 20, 2026. Supports: GA4 has an AI Assistant channel (ChatGPT, Gemini, Copilot and others) that excludes Google AI Overviews and AI Mode, which count under Organic Search.
  5. Introducing AI Performance in Bing Webmaster Tools Public Preview · Microsoft Bing Webmaster BlogOpened October 20, 2026. Supports: AI Performance reports citations in Copilot and Bing AI summaries; grounding-query data is a sample of overall citation activity.
How this page was made. Built on Rahil Jain's AI SEO method and drafted with AI research assistance. Every statistic links to the page we read it on (opened October 20, 2026 or nearby). Published October 20, 2026; last updated October 20, 2026. Chalk Labs sells AI SEO services, so we have a commercial interest in this topic. Spotted an error? Tell us and we will fix it.