Assistants like ChatGPT, Gemini, and Perplexity don’t show fixed results—they generate answers that vary with every run, every model, and every user.
“AI rank tracking” is a misnomer—you can’t track AI like you do traditional search.
But that doesn’t mean you shouldn’t track it at all.
You just need to adjust the questions you’re asking, and the way you measure your brand’s visibility.
In SEO rank tracking, you can rely on stable, repeatable rules:
- Deterministic results: The same query generally returns similar SERPs for everyone.
- Fixed positions: You can measure exact ranks (#1, #5, #20).
- Known volumes: You know how popular each keyword is, so you know what to prioritize.
AI breaks all three.
- Probabilistic answers: The same prompt can return different brands, citations, or response formats each time.
- No fixed positions: Mentions appear in passing, in varying order—not as numbered ranks.
- Hidden demand: Prompt volume data is locked away. We don’t know what people actually ask at scale.
And it gets messier:
- Models don’t agree. Even internal versions of the same assistant generate different responses to an identical prompt.
- Personalization skews results. Many AIs tailor their outputs to factors like location, context, and memory of previous conversations.
This is why you can’t treat AI prompts like keywords.
It doesn’t mean AI can’t be tracked, but that monitoring individual prompts is not enough.
Instead of asking “Did my brand appear for this exact query?”, the better question to ask is: “Across thousands of prompts, how often does AI connect my brand with this topic or category?”
That’s the philosophy behind Ahrefs Brand Radar—our database of millions of AI prompts and responses that helps you track directionally.
A major stumbling block when it comes to AI search tracking is that none of us know what people are actually searching en masse.
Unlike search engines, which publish keyword volumes, AI companies keep prompt logs private—that data never leaves their servers.
That makes prioritization tricky, and means it’s hard to know where to start when it comes to optimizing for AI visibility.
To move past this, we seed Brand Radar’s database with real search data: questions from our keyword database and People Also Ask queries, paired with search volume.
These are still “synthetic” prompts, but they reflect real world demand.
Our goal isn’t to tell you whether you appear for a single AI query, it’s to show you how visible your brand is across entire topics.
If you can see that you have great visibility for a topic, you don’t need to track hundreds of specific prompts within that topic, because you already understand the underlying probability that you’ll be mentioned.
By focusing on aggregated visibility, you can move past noisy outputs:
- See if AI consistently ties you to a category—not just if you appeared once.
- Track trends over time—not just snapshots.
- Learn how your brand is positioned against competitors—not just mentioned.
Think of AI tracking less like rank tracking and more like polling.
You don’t care about one answer, you care about the direction of the trend across a statistically significant amount of data.
You can’t track your AI visibility like you can track your search visibility. But, even with flaws, AI tracking has clear value.
Individual brand mentions in AI fluctuate a lot, but aggregating that data gives you a more stable view.
For example, if you run the same prompt three times, you’ll likely see three different answers.
In one your brand is mentioned, in another it’s missing, in a third a competitor gets the spotlight
But aggregate thousands of prompts, and the variability evens out.
Suddenly it’s clear: your brand appears in ~60% of AI answers.
Aggregation smooths out the randomness, outlier answers get averaged into the larger sample, and you get a better idea of how much of the market you actually own.
These are the same principles used in surveys: individual answers vary, but aggregate trends are reliable enough to act on.
They show you consistent signals you’d miss if you only focused on a handful of prompts.
The problem is, most AI tracking tools cap you at 50–100 queries—mainly because running prompts at scale gets expensive.
That’s not enough data to tell you anything meaningful.
With such a small sample, you can’t get a clear sense of your brand’s actual AI visibility.
That’s why we’ve built our AI database of ~100M prompts—to support the kind of aggregate analysis that makes sense for AI search tracking.
Studying how your brand shows up across thousands of AI prompts can help you spot patterns in demand, and test how your efforts on one channel impact visibility on the other.
Here’s what that looks like in practice, focusing on the example of Labubu (those creepy doll things that everyone has recently become obsessed with).
By combining TikTok data with Ahrefs Brand Radar, I traced how “Labubu” showed up across AI, social, search, and the wider web.
It made for an interesting timeline of events.
April: According to TikTok’s Creative Center, which allows you track trending keywords and hashtags, Labubu went viral on TikTok after unboxing videos took off in April.
May: Thousands of “Labubu” related search queries start showing up in the SERPs.
July: Search volume spikes for those same “Labubu” queries.
Also in July, web mentions for “Labubu” surge, overtaking market-leading toy Funko Pop.
August: Labubu crosses over into AI visibility, gaining mentions in Google’s AI Overviews in late August—overtaking another leading toy brand: Kaws.
Also in August, Labubu overtakes all other competitors in ChatGPT conversations.
This example shows that AI is part of a wider discovery ecosystem.
By tracking it directionally, you can see when and how a brand (or trend) breaks through into AI.
In all, it took four months for the Labubu brand to surface in AI conversations.
By running the same analysis on competitors, you can evaluate different scenarios, replicate what works, and set realistic expectations for your own AI visibility timeline.
AI variance shouldn’t stop you comparing your AI visibility to competitors.
The key is to track your brand’s AI Share of Voice across thousands of prompts—against the same competitors—on a consistent basis, to gauge your relative ownership of the market.
If a brand (e.g. Adidas) appears in ~40% of prompts, but a competitor (e.g. Nike) shows up in ~60% , that’s a clear gap—even if the numbers bounce around slightly from run to run.
Tracking AI search can show you the way your AI visibility is trending.
For example, if Adidas moves from 40% to 45% coverage, that’s a clear directional win.
Brand Radar supports this kind of longitudinal AI Share of Voice tracking.
Here’s how it works in five simple steps:
- Search your brand
- Enter your competitors
- Check your overall AI Share of Voice percentage
- Hit the “AI Share of Voice” tab to benchmark against your competitors
- Save the same prompt report and return to it to track your progress
Over time, these benchmarks show whether you’re gaining or losing ground in AI conversations.
A handful of prompts won’t tell you much, even if they are real.
But when you look at hundreds of variations, you can work out whether AI really ties your brand to its key topics.
Instead of asking “Do we appear for [insert query]?”, we should be asking “Across all the variations of prompts about this topic, how often do we appear?”
Take Pipedrive as an example.
CRM related prompts like “best CRM for startups” and “best CRM software for small business” account for 92.8% of Pipedrive’s AI visibility (~7K prompts).
But when you benchmark against the entire CRM market (~128K prompts), their overall share of voice drops to just 3.6%.
So, Pipedrive clearly “owns” certain CRM subtopics, but not the full category.
This brand of AI tracking gives you perspective.
It shows you how often you appear across subtopics and the wider market, but just as importantly, reveals where you’re missing.
Those gaps—the “unknown unknowns”—are opportunities and risks you wouldn’t have thought to check for.
They give you a roadmap of what to prioritize next.
To find those opportunities, Pipedrive can do a competitor gap analysis in three steps:
- Click “Others only”
- Study the prompt topics they’re missing in the AI Responses report
- Create or optimize content to claim some more of that visibility
AI results are noisy and synthetic prompts aren’t perfect, but that doesn’t stop them from revealing something crucial: how your brand is framed in the answers that do appear.
You don’t need flawless data to learn useful things.
The way AIs describe your brand—the adjectives they use, the sites they group you with—can tell you a lot about your positioning, even if the prompts are proxies and the answers vary.
- Are you labeled the “budget-friendly” option while competitors are framed as “enterprise-ready”?
- Do you consistently get recommended for “ease of use” while another brand is praised for “advanced features”?
- Are you mentioned alongside market leaders, or lumped in with niche alternatives?
These patterns reveal the narrative that AI assistants attach to your brand.
And while individual answers may fluctuate, those recurring themes add up to a clear signal.
For example, right now we have an issue with our own AI visibility.
Ahrefs’ positioning has shifted in the past year as we’ve added new features and evolved into a marketing platform.
But, AI responses still describe us primarily as an ‘SEO’ or ‘Backlinks’ tool.
By putting out consistent AI features, products, content, and messaging, our positioning is now beginning to shift on some AI surfaces.
You can see this when the red trend line (AI) overtakes the green (Backlinks) in the chart below.
Organic traffic is shrinking fast.
When Google’s AI Overview appears, clickthroughs to the top search results drop by about a third.
That means being named in AI answers is no longer optional.
AI assistants are already part of the discovery journey.
People turn to ChatGPT, Gemini, and Copilot for product recommendations, not just quick facts.
If your brand isn’t in those answers, you’re invisible at the exact moment decisions are made.
That’s why tracking AI visibility matters.
Even if the data is noisy, it shows whether you’re part of the conversation—or whether competitors are taking the spotlight.
In a perfect world, tracking AI visibility on a micro and macro level isn’t an either–or choice.
Micro tracking for high-stakes AI prompts
Micro tracking is about zooming in on the handful of queries that really matter to your business.
These might include:
- Branded prompts: e.g. “What is [Brand] known for?”
- Competitor comparisons: e.g. “[Brand] vs [Competitor]”
- Bottom-of-funnel purchase queries: e.g. “best [product] for [audience]”
Even though AI responses are probabilistic, it’s still worth monitoring these “make or break” queries where visibility or accuracy really matters.
Macro tracking for overall AI visibility
Macro tracking is about zooming out to understand the bigger picture of how AI connects your brand to topics and markets.
This approach is about monitoring thousands of variations to spot patterns, find new opportunities, and map the competitive landscape.
Most AI tools only handle the first mode, but Ahrefs’ Brand Radar can help you with both.
It lets you keep tabs on business-critical prompts while also surfacing the unknown unknowns.
And soon it’ll support custom prompts, so you can get even more granular with your tracking.
Looking at both levels helps you answer two questions: are you present where it counts, and are you strong enough to dominate the market?
Final thoughts
No, you’ll never track AI interactions in the same way you track traditional searches.
But that’s not the point.
AI search tracking is a compass—it will show if you’re headed in the right direction.
The real risk is ignoring your AI visibility while competitors build presence in the space.
Start now, treat the data as directional, and use it to shape your content, PR, and positioning.
