Part 2: How to Track Whether AI Is Citing You
This post is Part 2 of four in a series about getting cited by AI search engines like ChatGPT and Perplexity.
- Part 1: Answer Capsules: The Content Trait LLMs Cite Most
- Part 3: How to Audit Any Page for AI Citability
- Part 4: Page 1 on Google, Zero AI Citations
In Part 1 of this series, we covered the content traits that make LLMs cite your pages — answer capsules, link-free formatting, and original data. But writing quotable content is only half the problem. The other half is knowing whether it worked.
The core challenge of LLM tracking: There is no equivalent of Google Search Console for AI citations. No single dashboard shows you where ChatGPT, Perplexity, or Gemini are quoting your content.
That is the reality as of early 2026. And yet, tracking methods do exist. They are imperfect, manual in places, and evolving fast. This post covers what works today, what does not, and where the field is heading.
The framework we draw on here comes from Paul DeMott's research on Search Engine Land, which lays out a practical approach to LLM visibility tracking. We are building on his framework with our own observations from running an AI-powered SEO tool.
As with Part 1, this post uses the answer capsule technique throughout — every H2 is followed by a direct, self-contained answer before the supporting detail.
For the full GEO framework that connects tracking, content traits, and AI visibility strategy, see the Generative Engine Optimization guide.
Quick Navigation
- What You Can Track Today
- GA4 Referral Tracking
- The Manual Query Audit
- Share of Voice in LLM Responses
- The 62% Overlap Between SEO and LLM Citations
- Where Third-Party Mentions Matter
- Tools That Exist for LLM Tracking
- What You Cannot Track Yet
- A Practical Tracking Routine
- Frequently Asked Questions
What You Can Track Today
LLM visibility tracking currently combines three data sources: referral traffic in GA4, manual query audits, and emerging third-party tools.
None of these give you complete visibility. DeMott's research is direct about this: "nobody has complete visibility into LLM impact." Any vendor claiming otherwise is overpromising. But the combination of these three methods gives you a useful, actionable picture.
The good news is that you probably already have the first data source set up. If you are running GA4, you are already collecting LLM referral data — you just may not be looking at it.
GA4 Referral Tracking
GA4 captures traffic from ChatGPT, Perplexity, and other AI tools as referral sessions. You can see this in your traffic reports right now.
In GA4, go to Reports > Acquisition > Traffic acquisition and look at the Session source/medium dimension. Traffic from ChatGPT appears as chatgpt.com / referral. Perplexity shows as perplexity.ai / referral. These are real visits from people who read an AI-generated answer, saw your site cited, and clicked through.
This is the most reliable signal you have. It is actual traffic, not an estimate. If you are a Brass-SEO customer, your GA4 traffic report already includes this data — you can ask the AI chat to show you referral traffic from AI sources.
The limitation is that GA4 only captures clicks, not citations. If ChatGPT quotes your page but the user does not click through, you will never know. This means your GA4 referral data understates your actual AI visibility. It is a floor, not a ceiling.
The Manual Query Audit
The only way to see whether LLMs are citing you for specific queries is to ask them yourself and look at the results.
This sounds tedious, but it takes about 10 minutes for 5-8 queries. Open ChatGPT and Perplexity, type in the queries your target audience would ask, and note which domains appear in the answers. Are you there? Are your competitors? Which pages are getting cited?
We covered this process in detail in our post on finding AI citation gaps. The key insight from DeMott's research is to use a representative sample of 250-500 high-intent queries and run them regularly — daily or weekly — to track changes over time.
For most small businesses, 250 queries is overkill. Start with your top 10-20 target queries. The ones where you already rank in Google, or the ones where you know your content directly answers the question. Run them through ChatGPT and Perplexity once a month. Note which domains appear. That is your baseline.
Share of Voice in LLM Responses
Share of voice measures how often your brand appears in AI answers compared to competitors for a given set of queries.
This is the metric that DeMott's framework centers on. You pick a set of queries, run them through LLMs, and count how many times each brand is cited or mentioned. Your share of voice is your citation count divided by the total citations across all brands.
The distinction between "citations" and "mentions" matters. A citation is a linked source — the LLM explicitly references your page with a URL. A mention is a text reference — the LLM names your brand or product without linking to you. Both have value, but citations drive traffic.
This is similar to how traditional SEO share of voice works — instead of counting how many Google result positions you hold, you count how many AI answer slots you occupy.
Tracking share of voice manually is realistic for a small query set. For hundreds of queries, you need tooling.
The 62% Overlap Between SEO and LLM Citations
Brands ranking on Google's first page appeared in ChatGPT answers 62% of the time. Strong traditional SEO is still the best foundation for AI visibility.
This statistic from DeMott's research is important because it settles a common debate: do you need a completely separate strategy for LLMs, or does good SEO carry over? The answer is that SEO carries over significantly, but not completely.
The 62% overlap exists because many retrieval-augmented generation (RAG) systems — the architecture behind ChatGPT's Browse mode and Perplexity's search — pull from search engine results. If you rank well in Google, you are more likely to be in the pool of pages the LLM considers.
The remaining 38% is where content structure, formatting, and quotability make the difference. This is where the answer capsule technique from Part 1 becomes relevant. You need to rank well in Google AND format your content for LLM extraction.
Neither alone is sufficient. Both together give you the best odds.
Where Third-Party Mentions Matter
LLMs weight content that is referenced across multiple independent sources. Being mentioned on Reddit, review sites, and industry forums increases your citation likelihood.
DeMott's framework identifies several off-page targets that influence LLM visibility:
- Reddit and Quora discussions where your brand or product is mentioned in context
- Review websites and "best of" guides that list your product alongside competitors
- Wikipedia if your brand or topic has an entry
- Industry forums and communities where practitioners discuss tools in your category
This is not a new concept — it is essentially link building without cold emails applied to AI visibility. The difference is that LLMs weigh community mentions more heavily than traditional search engines do. A Reddit thread recommending your tool may not pass much PageRank, but it can significantly increase how often LLMs cite you.
The actionable takeaway: monitor where your competitors are mentioned that you are not. Those are your citation gap opportunities.
Tools That Exist for LLM Tracking
Several tools have emerged to automate LLM visibility tracking, though the space is early and no tool covers everything.
DeMott's research mentions Profound, Conductor, OpenForge, and Semrush as platforms that offer some form of LLM tracking. These tools generally work by running queries against LLMs at scale and recording which brands are cited, building share-of-voice dashboards over time.
We have not tested these tools ourselves, so we cannot endorse specific ones. What we can say is that any tool claiming "complete LLM visibility" should be treated with skepticism. The underlying data is inherently incomplete — LLM responses vary by session, user history, and model version. The best these tools can provide is a representative sample, not a census.
For small businesses, the manual approach (GA4 referrals + monthly query audits) is likely sufficient. The tooling becomes valuable when you are tracking hundreds of queries across multiple competitors and need to measure trends over time.
What You Cannot Track Yet
You cannot see how many times an LLM cited your page without the user clicking through. This is the biggest gap in current tracking.
When ChatGPT quotes your content in an answer and the user gets what they need without clicking your link, that is an invisible citation. You contributed to the answer. Your content was authoritative enough to be selected. But you have no way to measure it.
This is analogous to the early days of Google's AI Overviews, where featured snippets answered queries directly and reduced click-through rates. The exposure existed, but the traffic did not always follow.
Other things you cannot track yet:
- Training data citations. If your content was in ChatGPT's training data and influences answers without Browse mode, there is no way to know.
- Cross-session influence. If a user sees your brand cited by an LLM, then Googles you later, that touchpoint is invisible.
- Model-specific behavior. Different models may cite you at different rates for the same query. Testing one model does not tell you about the others.
The field is moving fast. These gaps will likely narrow over the next 12-18 months as tracking tools mature and LLM platforms potentially release their own analytics.
A Practical Tracking Routine
A monthly tracking routine takes 30 minutes and gives you a usable baseline for LLM visibility.
Here is what we recommend for small businesses:
Weekly (5 minutes): Check GA4 for chatgpt.com / referral and perplexity.ai / referral traffic. Note any changes.
Monthly (25 minutes):
- Pick your top 10 target queries — the ones your best content answers
- Run each through ChatGPT and Perplexity
- Record which domains are cited in each answer (including yours)
- Note any queries where competitors appear and you do not
- Compare to last month's results
Quarterly: Review which posts are receiving AI referral traffic in GA4. Cross-reference with the content structure changes you have made (answer capsules, original data). Use your Google Search Console data to see if the same posts are also gaining in traditional search.
This is not a complex analytics setup. It is a notebook (or spreadsheet) and 30 minutes. The goal is to build a trendline over time, not to get a perfect snapshot on any given day.
For a more recent quarterly testing method across four AI engines with a competitive scoring framework, see How to Test Your Brand's AI Search Visibility.
Frequently Asked Questions
How do I see AI referral traffic in GA4?
Go to Reports > Acquisition > Traffic acquisition and filter by Session source. Look for chatgpt.com, perplexity.ai, claude.ai, or gemini.google.com as referral sources. These show actual clicks from users who saw your content cited in an AI answer.
How often should I check my LLM visibility?
Weekly GA4 checks take 5 minutes and catch any sudden changes. Monthly manual query audits take 25 minutes and give you a share-of-voice baseline. Quarterly reviews help you spot longer-term trends.
Do I need paid tools to track LLM visibility?
Not for most small businesses. GA4 (free) plus manual query audits give you a solid foundation. Paid tools become valuable when you are tracking hundreds of queries and need automated trend reporting.
Does ranking well in Google automatically get me cited by LLMs?
Partially. Research shows a 62% overlap — brands on Google's first page appeared in ChatGPT answers 62% of the time. But the remaining 38% depends on content structure, formatting, and third-party mentions.
How long until I see results from LLM optimization?
LLMs incorporate new content faster than Google — days rather than weeks. But building sustained visibility requires 6-12 months of consistent effort, similar to traditional SEO.
What is the relationship between this post and Part 1?
Part 1 covers what content traits get you cited by LLMs — the answer capsule technique, link-free formatting, and original data. This post (Part 2) covers how to measure whether those changes are working. Both posts use the answer capsule technique in their own structure.
This is Part 2 of our LLM Visibility series. Part 1 covers content traits. Part 3 covers auditing. Part 4 is a before-and-after case study.
The tracking framework in this post draws on Paul DeMott's research published on Search Engine Land in October 2025. The 62% SEO-LLM overlap statistic and the share-of-voice methodology are from that research. Our recommendations for small business tracking routines are based on our own experience running Brass-SEO.