ChatGPT, Gemini & Claude Lead AI Visibility, Is It Time to Stop Tracking Perplexity?

There is a growing debate in the SEO community about which AI platforms actually deserve our attention. Recently, Ross Hudgens of Siege Media suggested that because Perplexity's market share is dipping, marketers should simply stop tracking it. The logic is straightforward: if a platform isn't driving significant traffic, including it in your visibility scores only distorts the data. This connects with Are They Measuring the Right Things? when the same signal needs a clearer operating decision.

While the question is the right one to ask, I am not convinced that deleting the platform from your dashboard is the right answer. The AI search landscape is consolidating, but it is doing so in a way that suggests a tiered ecosystem rather than a simple leaderboard. If we move too quickly to ignore the "smaller" players, we risk repeating the mistakes of the early search era.

The Danger of Declaring a Market Settled

This conversation reminds me of a specific moment in 2002 at a Search Engine Strategies conference in Boston. At the time, the dot com bubble had burst, and many agencies were tracking rankings across a dozen or more different search engines. A common argument at the time was that we should simplify everything and focus only on the five giants: Yahoo, Excite, Lycos, AltaVista, and Ask Jeeves.

The glaring omission in that list was Google. While not yet the undisputed incumbent, Google was gaining ground at a pace that would soon make the others irrelevant. That experience taught me to be cautious whenever someone suggests a market is "settled."

The real question for an SEO professional isn't just who is the biggest today. It is which platforms are growing fast enough or hold enough strategic importance that ignoring them would leave you blind to the next major shift in user behavior.

Expert Interpretation: The tradeoff here is between operational efficiency (tracking fewer things) and strategic foresight (tracking emerging threats). The decision to drop a platform should be based on a lack of growth and lack of strategic relevance, not just a lower current volume compared to a giant like OpenAI.

Analyzing the Decline of Perplexity

To be fair, the data supporting the argument for removing Perplexity is compelling. StatCounter data from 2026 shows a sharp decline in referral share. In June 2026, Perplexity held about 7.91% of the AI chatbot referral share, putting it neck and neck with Gemini. By August, that number dropped to 4.31%, while Gemini climbed to 10.9%.

When you look at total web visits via Similarweb data from May 2026, the gap is even wider. ChatGPT dominated with 53.9% of visits, followed by Gemini at 27.9% and Claude at 9.2%. Perplexity and Copilot trailed far behind, both sitting at around 1.3%.

This creates a legitimate problem with weighting. If your tracking tool treats a citation on Perplexity as equal to a citation on ChatGPT, your "average visibility score" becomes a vanity metric. A high score on a low traffic platform can mask a failure on a high traffic one.

Expert Interpretation: We must distinguish between referral share (traffic sent to sites) and usage share (how many people use the tool). A platform might have low referral traffic but high influence on how users perceive a brand. The risk is treating a "low traffic" signal as a "zero value" signal.

A Market of Three Distinct Powerhouses

The most significant trend of 2026 isn't the fall of one player, but the rise of a multi polar market. We are seeing a redistribution of attention. ChatGPT's share of AI chatbot web traffic dropped from over 76% a year prior to roughly 52.7% by May 2026. Meanwhile, Gemini jumped from 9% to 27.3%, and Claude grew from 1.6% to 8.9%.

However, this isn't a zero sum game where ChatGPT is dying. OpenAI reported over one billion active users across its products by July. Google's Gemini app also passed the one billion monthly user mark in August. Claude has carved out a different niche entirely, focusing on enterprise and developer workflows. Anthropic noted that over 100,000 customers were utilizing Claude via Amazon Bedrock as of April.

These three aren't just competing for the same user; they have different distribution advantages. ChatGPT has the direct consumer lead, Gemini has the Android and Google ecosystem, and Claude has the professional and enterprise foothold.

Expert Interpretation: This suggests that "AI Visibility" is not a single metric. Depending on your business model, one of these three may be significantly more important than the others. A consumer app developer should prioritize ChatGPT, while a B2B software provider should look closer at Claude. A useful companion note is SEO. Don’t Let Claude Do SEO., because it looks at a nearby part of the same system. The same pattern also shows up in New Data Suggests, where the practical question is how the signal becomes visible.

The Unique Position of Google AI Search

Tracking becomes conceptually difficult when we bring Google's integrated AI features into the mix. I believe Google AI Overviews and AI Mode should not be treated as just another LLM in a list. They are an entirely different category of visibility.

The scale is simply too large to ignore. Google reported that AI Overviews reached over 2.5 billion users per month by June, and AI Mode surpassed one billion. By May 2026, AI Overviews appeared in 43% of US Google searches, a massive jump from 15% the previous year. AI Mode visits also surged, growing from 126 million in June 2025 to 279 million in May 2026.

If your goal is to understand how people discover information, you cannot treat Google's AI layer as a niche chatbot. It is the primary interface for the world's most dominant search engine.

Expert Interpretation: The tradeoff here is between simplicity and accuracy. It is easier to put everything in one "AI" bucket, but that obscures the fact that AI Overviews are a modification of traditional search, whereas ChatGPT is a destination. You should track these as two separate behaviors: "AI enhanced search" and "Conversational AI."

Why Traffic Data Doesn't Tell the Whole Story

Claude is the perfect example of why relying solely on consumer traffic data is dangerous. If you only look at web visits, Claude looks small. But for B2B companies, Claude is becoming a primary tool for professional work.

The enterprise adoption is staggering. PwC is deploying Claude Code and Cowork to 30,000 professionals. TCS is making it available to 50,000 employees across 56 countries, and Cognizant has trained over 30,000 associates on the platform. Anthropic reported that over 1,000 business customers were spending more than $1 million annually as of April 2026.

For a B2B marketer, a citation in Claude might be worth ten times more than a citation in a consumer facing bot because the user is likely a high value decision maker in a corporate environment.

Expert Interpretation: This highlights the necessity of aligning your tracking portfolio with your customer persona. A B2C retailer and a cybersecurity vendor cannot use the same AI tracking list because their audiences live in different ecosystems.

A Three Tier Framework for AI Visibility

Rather than simply deleting platforms, I suggest organizing your AI visibility measurement into three distinct tiers based on strategic significance.

Tier One: Scale and Strategy. These are the platforms with massive reach and independent ecosystems. Track ChatGPT, Gemini, and Claude. Crucially, track them separately. Do not average them into a single "LLM score," as their audiences and use cases differ too much.

Tier Two: Ecosystem Integration. This tier covers AI embedded in existing software. Google AI Overviews and AI Mode belong here. Microsoft Copilot also fits this category, especially for those targeting the Microsoft 365 ecosystem, which reaches 900 million monthly active users through various AI features.

Tier Three: Emerging and Specialized. This is where Perplexity and other niche or new entrants live. These should be monitored, but they should not carry the same weight as Tier One or Two in your reporting.

Expert Interpretation: The decision here is about resource allocation. You spend the most time optimizing for Tier One and Two, while using Tier Three as an "early warning system" to spot new trends before they hit the mainstream.

Avoiding the Trap of the Average Score

I agree with the core of the argument that aggregate scores can be misleading. If a brand has a 90% citation rate on Perplexity but only 30% on ChatGPT, a simple average might show a healthy 60% visibility. But if Perplexity only drives 1% of the traffic and ChatGPT drives 50%, that 60% is a lie.

The goal isn't to find an average; it is to find where the people who matter to your business are actually encountering your brand. To do this, you need to synthesize three different data points:

Audience Exposure: Where is your specific target audience spending their time? Visibility: Where are you being mentioned, cited, or linked? Trigger Prompts: Which specific queries are causing the AI to mention your brand?

Expert Interpretation: The mistake most marketers make is focusing on the "what" (the score) instead of the "who" (the audience). You must weight your visibility data by the actual traffic or user value of the platform.

The Verdict on Perplexity

Should you remove Perplexity from your tracker? No. But you should absolutely remove it from a position of equal weighting.

If Perplexity represents a tiny fraction of your AI referrals, giving it 25% of your visibility score is logically indefensible. However, if it represents 5% or more of your referrals, or if your specific industry sees a disproportionate amount of Perplexity usage, deleting it would be a mistake. You would be erasing a competitive signal that could be vital for your specific vertical.

The lesson from the early 2000s is that the danger isn't in tracking too many things; it is in assuming that today's leaders are the only ones who matter tomorrow. The market is shifting, and while the giants are currently dominating, the most successful SEOs will be those who weight their data accurately without closing their eyes to the fringes.

Expert Interpretation: The final decision comes down to a simple audit: look at your referral traffic. If a platform is driving meaningful leads, keep it. If it isn't, move it to Tier Three. Never delete a signal entirely just because it is currently small.