A debate kicked off on LinkedIn is forcing marketers to rethink how they score AI visibility. Ross Hudgens, CEO of Siege Media, argued bluntly that everyone should pull Perplexity out of their LLM trackers, because its shrinking share distorts the picture of where visibility actually pays off.
Search Engine Journal pushed back — not on the diagnosis, but on the cure. The market is consolidating. It just isn’t finished.
The numbers behind the argument
StatCounter’s worldwide AI chatbot referral data shows the shift clearly. In June 2026, Perplexity sat at 7.91% of AI chatbot referral share, effectively level with Gemini at 7.94%. By August, Perplexity had slid to 4.31% while Gemini climbed to 10.9%.
Usage share tells a similar story. Similarweb’s May 2026 data across seven major AI assistants put ChatGPT at 53.9% of worldwide web visits, Gemini at 27.9%, Claude at 9.2%, DeepSeek at 4.1%, Grok at 2.4%, and Perplexity and Copilot both at 1.3%.
So yes — if your dashboard gives Perplexity the same weight as ChatGPT, one freak result can swing your whole score.
The real 2026 story is a second superpower
The headline isn’t Perplexity fading. It’s Gemini arriving. ChatGPT’s share of AI chatbot web traffic fell from 76.4% a year earlier to roughly 52.7% in May 2026, while Gemini jumped from about 9% to 27.3% and Claude went from 1.6% to 8.9%.
Nobody is dying, though. OpenAI reported more than 900 million weekly active users in February and said in late July its models reach over a billion active users across products. Google said in August that the Gemini app passed 1 billion monthly users. Anthropic told investors in August that annualised revenue had crossed $65 billion, after saying in April that more than 100,000 customers were running Claude on Amazon Bedrock.
Three platforms, three completely different distribution engines: consumer habit, Google’s ecosystem, and enterprise workflow. That’s an oligopoly, not a leaderboard.
Google’s AI layer is its own category
Here’s where most trackers get sloppy. Google reported in June that AI Overviews reach more than 2.5 billion users monthly and AI Mode had passed 1 billion monthly users, with AI Mode queries more than doubling every quarter since launch. Similarweb data reported by TechCrunch showed AI Overviews appearing in 43% of US Google searches by May 2026, up from 15% a year earlier, while AI Mode visits grew from 126 million in June 2025 to 279 million in May 2026.
That isn’t “another LLM.” It’s an AI layer inside the dominant search ecosystem — and it deserves its own row in your reporting.
Why usage share alone will mislead B2B teams
Claude looks small on consumer traffic charts and enormous inside enterprises. PwC said it would roll out Claude Code and Cowork across its global workforce and certify 30,000 professionals. TCS announced plans to make Claude available to 50,000 employees across 56 countries. Cognizant says more than 30,000 associates have completed Claude training. Anthropic reported more than 1,000 business customers spending over $1 million annually as of April 2026.
Translation for growth teams: a D2C retailer, a SaaS company and a cybersecurity vendor should not be tracking the same portfolio of AI platforms.
The three-tier tracking framework
- Tier one — scale plus strategic weight: ChatGPT and Gemini, tracked separately rather than blended. Add Claude if you sell to enterprise, developer or professional audiences.
- Tier two — AI inside established ecosystems: Google AI Overviews and AI Mode, reported apart from the chatbots. Copilot belongs here for Microsoft 365-heavy organisations; Microsoft says its first-party Copilot family has over 150 million monthly active users, with AI features across its products reaching 900 million.
- Tier three — emerging and specialised: Perplexity, Grok, DeepSeek and friends. Monitor for unusual citations, referral spikes and growth. Don’t average them into your headline number.
Kill the vanity average, not the data point
The strongest part of the critique is about false precision. Say you score 40% citation rate on ChatGPT, 35% on Gemini, 30% on Claude and 90% on Perplexity. The average is a tidy 48.75% — and it’s meaningless if Perplexity drives 1% of your AI referrals.
Better model: layer three datasets. Audience exposure (usage and visits by platform), visibility (mentions, citations, linked URLs and the prompts triggering them), and business impact (AI referrals mapped to leads, sales and revenue).
The verdict is a de-weighting, not a deletion. As SEJ notes, if Perplexity is 1% of your AI referrals, letting it own a third of your score is indefensible — but if it’s above 5%, or if your vertical over-indexes there, cutting it destroys a competitive signal.
The historical warning is sharp: back in 2002, the “five engines that mattered” list left Google off entirely. Weight by reality. Keep one eye on the outsiders.
Source: Search Engine Journal
