Search Engine Journal argues Perplexity should be de-weighted in AI visibility tracking after its referral share fell to 4.31 percent in August

Search Engine Journal published a piece by Greg Jarboe on 2026-09-09 responding to a LinkedIn call from Ross Hudgens, CEO of Siege Media, to drop Perplexity from AI visibility trackers. Jarboe agrees Perplexity has lost ground, disagrees with removing it, and lays out the market-share numbers behind the argument.

The referral share numbers, from StatCounter

Jarboe cites StatCounter worldwide AI chatbot referral data showing Perplexity at 7.91 percent in June 2026, essentially tied with Gemini at 7.94 percent. By August, Perplexity had fallen to 4.31 percent, and Gemini had risen to 10.9 percent. That is the shift Hudgens is reacting to.

A separate Similarweb picture for platform size

Jarboe quotes Similarweb's May 2026 data on worldwide web visits across seven major AI assistants: ChatGPT 53.9 percent, Gemini 27.9 percent, Claude 9.2 percent, DeepSeek 4.1 percent, Grok 2.4 percent, Perplexity 1.3 percent, and Copilot 1.3 percent. He also notes ChatGPT's share of AI chatbot web traffic fell from 76.4 percent a year earlier to 52.7 percent in May 2026, while Gemini rose from a Similarweb-reported 9 percent a year earlier to 27.3 percent and Claude from 1.6 percent to 8.9 percent.

Why Claude looks small and matters anyway

Jarboe uses Claude to argue that consumer usage share alone is a bad ranking. Anthropic, per its own announcements he cites, has enterprise-distribution deals with PwC (planning to train and certify 30,000 professionals on Claude Code and Cowork), TCS (Claude for 50,000 employees across 56 countries), and Cognizant (more than 30,000 associates completed Claude training). Anthropic said in April 2026 that more than 1,000 business customers were spending more than $1 million a year with it.

Where Google fits

Jarboe argues Google AI Overviews and AI Mode should not be treated as one more LLM in the same list. Google said AI Overviews were reaching more than 2.5 billion users a month and AI Mode had passed 1 billion monthly users. Similarweb data reported by TechCrunch put AI Overviews on 43 percent of United States Google searches by May 2026, up from 15 percent a year earlier. Microsoft says its first-party Copilot family has more than 150 million monthly active users.

The tracking model Jarboe proposes

Three tiers, not one leaderboard. Tier one: ChatGPT and Gemini tracked separately as the platforms with scale and strategic weight, plus Claude for business-to-business and enterprise audiences. Tier two: Google AI Overviews and AI Mode as a separate search-discovery layer, with Microsoft Copilot alongside for organisations already deep in Microsoft 365. Tier three: emerging platforms, including Perplexity, Grok and DeepSeek, monitored for unusual visibility, referrals, citations and growth rather than weighted equally.

The prescription is a warning against a familiar shape in a marketing dashboard. Average four platforms and a brand with 40 percent visibility on ChatGPT, 35 on Gemini, 30 on Claude and 90 on Perplexity gets a headline number of 48.75 percent, which reads well and describes almost nothing about where the reader will actually see the brand. Weighting each platform against real referral traffic replaces that with a number that follows the audience.

Source

• ChatGPT, Gemini & Claude Lead AI Visibility, Is It Time To Stop Tracking Perplexity?, Greg Jarboe, Search Engine Journal, 2026-09-09