Brands Are Tracking AI Visibility, but Are They Measuring the Right Things?

There is a sudden, intense pressure on marketers to figure out how their brands appear in AI responses. We have moved past the curiosity phase and into the measurement phase. If you can't prove your brand is being mentioned by ChatGPT or appearing in a Google AI Overview, it feels like you are losing a game you don't fully understand. This connects with Which Brands Are Vanishing from AI Search? when the same signal needs a clearer operating decision. A useful companion note is Two Ways Brands Appear in AI Search, because it looks at a nearby part of the same system.

The problem is that we are often chasing a number without knowing what that number actually represents. When a "visibility score" moves, the instinct is to either celebrate or panic. But before you change your entire content strategy based on a dashboard, it is worth questioning whether the metrics you are tracking actually correlate with business growth.

The Critical Difference Between Mentions and Citations

Many AI visibility tools bundle different types of presence into a single score, but a mention and a citation are fundamentally different events. A mention occurs when the AI simply says your brand name in a response. A citation is when the AI provides a clickable URL as a source.

This distinction matters because you can have one without the other. You might be mentioned as an authority in a response, but if there is no link, there is no direct path for the user to reach your site. Conversely, your page might be used as a source citation, but the AI might not even mention your brand name in the prose of the answer.

The gap between retrieval and citation is even wider. Research from Ahrefs into 1.4 million ChatGPT prompts revealed that while Reddit URLs were retrieved frequently, they were actually cited in only 1.93% of cases. This suggests that AI models are reading and processing vast amounts of data that never actually make it into the final, visible response.

Expert Interpretation: The tradeoff here is between "brand awareness" and "traffic acquisition." A mention is a branding win; a citation is a performance win. If you only track a combined score, you won't know if you are becoming a household name that no one clicks on, or a utility source that no one remembers. Inspect your data to see if your "visibility" is actually driving clicks or just providing free training data for the LLM.

Why Traditional Search Rankings Still Carry Weight

It is tempting to think that AI search is a completely new ecosystem that ignores the old rules of SEO. The data suggests otherwise. Traditional organic rankings still play a significant role in who gets cited in AI Overviews.

In an analysis of 4 million AI Overview URLs across 863,000 keywords, about 37.1% of the cited URLs were also in the organic top 10 for that same query. Another 26.2% were ranked between positions 11 and 100. While a large chunk (36.7%) didn't make the top 100, the correlation with traditional rankings remains strong.

The reason some low ranking pages still get cited is due to a process called "query fan out." When Google generates an AI Overview, it doesn't just look at the primary keyword. It breaks the main query into several related sub queries. This allows a page that is an expert on a specific sub topic to be cited, even if it doesn't rank well for the broad, head term.

Expert Interpretation: This means traditional SEO is not dead; it has just become more fragmented. The decision you need to make is whether to keep fighting for the top 3 spots on high volume head terms or to build a "web" of deep, topical authority that captures these sub queries. The risk of ignoring traditional SEO is that you lose the foundation that feeds a large portion of AI citations.

The Misconception of Technical Signals like Schema

There is a common belief in the SEO community that adding specific technical markers, like JSON LD schema markup, will act as a "signal" to AI models to cite your content more often. The reality is more nuanced and less predictable.

Data shows that pages cited by AI are nearly three times more likely to have JSON LD than those that aren't. On the surface, this looks like a cause and effect relationship. However, when researchers tracked 1,885 pages that actually added schema between August 2025 and March 2026, the results were surprising. Adding schema did not significantly increase citations for ChatGPT or AI Mode.

In fact, citations for AI Overviews actually fell by 4.6% for the group that added schema. While this decrease was small, it was statistically significant. It is unclear if the schema caused the drop or if it was a coincidence, but it proves that technical "hacks" are not a guaranteed lever for AI visibility.

Expert Interpretation: This highlights a classic correlation versus causation trap. High quality sites likely use schema because they are professional and thorough, and those same sites are cited because they are high quality, not because of the schema itself. Do not prioritize technical markup over content depth in the hopes of "tricking" an AI into citing you.

Turning AI Data Into Actionable Strategy

If the scores are volatile and the signals are confusing, how do you actually use this data? The first step is acknowledging that AI responses are probabilistic. The same prompt can yield different brands and different citations every time you run it.

To get a real signal, you cannot rely on a few manual searches. You need to analyze large sets of prompts to find a pattern. If you see a shift in your score, you must first determine if it is a real market change or simply a result of the tool updating its prompts or switching the underlying model it is tracking.

Once you have a reliable baseline, you can start asking specific questions:

If mentions are dropping but your rankings are steady, are competitors stealing your "share of voice" in the AI's mind? If citations are dropping but you still rank well, is your content failing to answer the specific sub queries the AI is generating through fan out? Is the decline happening across all platforms, or is it isolated to one, such as ChatGPT or Google?

Expert Interpretation: The tradeoff here is between speed and accuracy. It is fast to look at a single score, but it is accurate to analyze prompt sets. The decision to make is whether you are reporting "vanity metrics" to stakeholders or performing a genuine gap analysis of your brand's positioning.

The Danger of the "Visibility Score"

The most dangerous mistake a marketer can make is treating AI visibility as a proxy for success. A brand can see its visibility score climb while its actual traffic plummets.

Research comparing data from December 2023 to December 2025 shows a stark reality. For keywords where an AI Overview appeared, the clickthrough rate (CTR) for the top organic result was 58% lower than it would have been without the AI Overview. While CTR was dropping across the board during this period, the AI presence added a significant extra layer of decline.

This happens because AI search is designed to satisfy the user's intent directly on the search page. Being featured in the answer is a win for brand awareness, but it is often a loss for traffic. You are essentially providing the answer for free, removing the need for the user to click through to your site.

To avoid this trap, you must combine visibility data with actual performance data. Tools like Google Search Console's AI performance reports are essential here. They allow you to see the difference between being "seen" and being "visited."

Expert Interpretation: We are moving from an era of "Traffic Acquisition" to an era of "Influence Management." If your goal is purely clicks, a high AI visibility score might actually be a warning sign of declining traffic. You must decide if your current KPI is "clicks to site" or "brand authority in the AI's knowledge graph," as the strategies to optimize for each are fundamentally different.