How to Track AI Citations for Your Brand

AI citation tracking is the measurement layer of GEO. Without it, you are optimizing blind — you do not know whether your content changes, schema improvements, or digital PR efforts are actually increasing your visibility in AI answers. This guide walks through building a manual citation tracking system that works across ChatGPT, Perplexity, Google AI Overviews, and Gemini.

Step 1: Build your target query list

Your tracking system is only as good as the queries you track. Build a list of 20-50 queries across four categories:

• Brand queries (5-10) — "what is [your brand]", "[your brand] reviews", "is [your brand] legit"
• Category queries (10-15) — "best [your service]", "top [your service] tools", "[your service] alternatives"
• Comparison queries (5-10) — "[your brand] vs [competitor]", "[competitor A] vs [competitor B]"
• Informational queries (5-15) — "how does [your service] work", "what is [your industry term]"

Source queries from:

• Google Search Console (queries driving traffic)
• People Also Ask boxes
• Customer support tickets and sales calls
• Competitor pages and keyword tools

Record each query in a spreadsheet. This list stays constant — you test the same queries each tracking run to measure change over time. See how to run a GEO audit for the full query selection process.

Step 2: Set up a tracking spreadsheet

Create a spreadsheet with this structure:

| Query                          | Engine        | 2026-09-01 | 2026-09-08 | 2026-09-15 |
|--------------------------------|---------------|------------|------------|------------|
| best SEO consultant London     | ChatGPT       | Absent     | Absent     | Cited      |
| best SEO consultant London     | Perplexity    | Competitor | Competitor | Competitor  |
| best SEO consultant London     | AI Overviews  | Absent     | Absent     | Mentioned   |
| best SEO consultant London     | Gemini        | Absent     | Absent     | Absent      |
| what is GEO                   | ChatGPT       | Absent     | Cited      | Cited       |
| what is GEO                   | Perplexity    | Absent     | Cited      | Cited       |

Columns:

• Query — the target query
• Engine — ChatGPT, Perplexity, AI Overviews, Gemini
• Date columns — one per tracking run
• Status values — Cited (with link), Mentioned (no link), Absent, Competitor (competitor cited)

Add a summary tab that calculates citation rate and share of voice per tracking date. See what is GEO for why this metric matters.

Step 3: Run queries across all four AI engines

For each query, test it in all four engines:

Record the status for each. This takes 1-2 hours for 30 queries. It is manual, but it is the most reliable method — AI engine APIs do not provide citation data, and automated tools are still maturing. See Profound vs Otterly for emerging tools in this space.

Step 4: Track competitor citations alongside your own

For each query where you are absent, record which competitor is cited. Choose 3-5 key competitors and track them across the same queries.

| Query                      | Your Brand | Competitor A | Competitor B | Competitor C |
|----------------------------|------------|--------------|--------------|--------------|
| best SEO consultant London | Absent     | Cited        | Mentioned    | Absent       |
| what is GEO               | Cited      | Absent       | Absent       | Cited        |

This gives you share of voice data — what percentage of AI citations in your space go to you vs competitors. Share of voice is more meaningful than raw citation count because it accounts for the total competitive landscape.

Step 5: Calculate citation rate and share of voice

Two key metrics:

Citation rate = (queries where you are cited / total queries) x 100

Example: You are cited in 8 out of 30 queries.
Citation rate = (8 / 30) x 100 = 26.7%

Share of voice = (your citations / total citations across all tracked brands) x 100

Example: Across 30 queries, there are 45 total citations.
Your brand: 8 citations. Competitor A: 15. Competitor B: 12. Competitor C: 10.
Your share of voice = (8 / 45) x 100 = 17.8%

Track these metrics over time. A rising citation rate and rising share of voice mean your GEO efforts are working. A falling rate means something changed — investigate and fix. See SEO vs AEO vs GEO for how these metrics complement traditional SEO.

Step 6: Schedule regular tracking runs

Run the full tracking set on a regular cadence:

• Weekly — best for active GEO campaigns where you are making changes and want to see impact
• Biweekly — good for steady-state monitoring
• Monthly — too infrequent for most brands; you will miss shifts

AI engine behavior changes with:

• Model updates (new GPT versions, new Gemini versions)
• Feature changes (new search modes, citation format changes)
• Index updates (Bing reindexing, Google algorithm updates)

Weekly tracking catches these shifts early. Block 1-2 hours on your calendar each week for the tracking run.

Step 7: Connect citation data to GEO actions

Tracking without action is wasted effort. Use the data to drive decisions:

• Citation rate drops for a specific query — Investigate: did a competitor publish better content? Did your page change? Did the AI engine update? Fix and re-test.
• Citation rate rises after content restructuring — You have evidence that your GEO work is effective. Double down on the same approach for other queries.
• Consistently absent for category queries — The gap is likely entity authority or content extractability. See how to get cited by ChatGPT and how to optimize content for AI Overviews.
• Cited by Perplexity but not ChatGPT — Check Bing visibility (ChatGPT uses Bing). See ChatGPT Search vs Perplexity for why they differ.
• Cited in AI Overviews but not Perplexity — Check whether your content is crawlable by Perplexity's bot and whether it is factual and well-sourced. See Perplexity vs Google AI Overviews.

See GEO consulting for help building a tracking system and acting on the data, and the glossary for terminology.

Common mistakes

• Tracking too few queries — 5 queries is a spot check, not a tracking system. You need 20-50 queries to see meaningful patterns. Too few queries leads to false confidence or false alarm from a single query change.
• Only tracking one AI engine — Citation rates vary significantly across engines. Being cited in Perplexity does not mean you are cited in ChatGPT. Track all four. See ChatGPT Search vs Perplexity and Google AI Overviews vs Gemini for why they differ.
• Not tracking competitors — Your citation rate in isolation is less meaningful than your share of voice. If your rate goes from 20% to 25% but a competitor went from 30% to 50%, you are losing ground. Always track competitors.
• Tracking monthly or quarterly — AI engine behavior changes too fast for monthly tracking. By the time you notice a drop, it may have been happening for weeks. Track weekly or biweekly. See how to run a GEO audit for the quarterly full-audit process.
• Not connecting tracking data to actions — Tracking without action is just data collection. Every tracking run should produce a list of actions: queries to investigate, content to restructure, schema to fix. See GEO consulting and AEO consulting for help acting on citation data.