ChatGPT’s Share of B2B AI Traffic Fell to 63%. The Winner Isn’t Google.

The chart on every AI search slide deck right now shows ChatGPT owning roughly three-quarters of the market and everyone else fighting over scraps. StatCounter puts ChatGPT at 76.85% of AI chatbot referrals as of April 2026, down from 84.21% a year earlier (StatCounter). Slipping, but still dominant. If you sell to consumers, that chart is roughly true.

If you sell to other businesses, that chart is lying to you.

In B2B, ChatGPT’s grip came loose far faster than the aggregate number admits. Goodie’s 2026 AI Search Traffic Report tracked referral data from a panel of business-facing brands across two windows: May to August 2025, then March to April 2026. ChatGPT’s share of those B2B AI referrals fell from 89.1% to 62.6%, a drop of more than 26 points in eight months (Goodie).

A quarter of the channel changed hands in less than a year. And the brand that took the biggest piece of it is one most B2B marketing teams aren’t tracking at all.

The winner is Claude, not Google

When marketers picture ChatGPT losing ground, they picture Google clawing it back with Gemini and AI Overviews. That’s the SEO-era reflex: the two giants trade blows, everyone else is noise.

Not this time. In Goodie’s B2B panel, Claude went from 1.4% of referrals to 18.5%, a 17-point gain that vaulted it past Gemini, past Perplexity, into clear second place behind ChatGPT. Gemini and Perplexity grew too, but at a fraction of the pace.

Platform	B2B AI referral share, mid-2025	B2B AI referral share, early 2026	Change
ChatGPT	89.1%	62.6%	−26.5 pts
Claude	1.4%	18.5%	+17.1 pts
Gemini	2.4%	10.6%	+8.2 pts
Perplexity	3.1%	7.3%	+4.2 pts
Copilot	3.2%	4.0%	+0.8 pts

Source: Goodie 2026 AI Search Traffic Report. Figures are brand-averaged shares of measurable B2B AI referrals.

Read the bottom of that table and the whole “Big Two” frame falls apart. Four platforms now account for nearly all measurable B2B AI referrals, and the fastest mover among them is the one that doesn’t run ads, doesn’t have a consumer search product, and barely registers on the charts most teams are looking at.

Why the headline chart misses this

This is the part worth slowing down for, because it explains how a 26-point swing hid in plain sight.

The market-share numbers everyone quotes are weighted by total web traffic, and total web traffic is overwhelmingly consumer. When SE Ranking measured Claude across its full dataset, Claude came out at just 1.40% of all AI-referred traffic, dwarfed by ChatGPT’s 78.23% (Search Engine Journal). Look only at that number and Claude looks like a rounding error.

But the same dataset shows Claude was the fastest-growing AI traffic source of all, climbing 386% between January and April 2026 while ChatGPT grew 1.5%. Small base, steep curve. And the curve is steepest exactly where B2B buyers live, which is why Claude reads as 1% of the consumer web and 18% of B2B referrals at the same time. Both numbers are correct. They’re measuring different populations.

The blended chart averages your buyers together with everyone asking an AI to write a birthday poem or summarize a recipe. For a B2B brand, that average isn’t a useful approximation. It’s the wrong question answered precisely.

Claude is a work tool, and work is where your buyers are

The reason B2B fragmented faster than the consumer market isn’t a mystery. It’s about what Claude is used for.

In May 2026, the Ramp AI Index (which tracks real corporate spending across more than 50,000 U.S. businesses) showed Anthropic passing OpenAI in business adoption for the first time, at 34.4% of business AI spend versus 32.3% (TechCrunch). Anthropic roughly quadrupled its business footprint over the year while OpenAI’s barely moved. The engine behind it was Claude Code and Claude’s pull inside engineering and knowledge-work teams.

Follow the logic to its end. The person evaluating your software is a developer, a technical lead, an analyst, an ops manager. They already have Claude open all day for code, docs, and research. When they need to compare vendors or sanity-check a shortlist, they ask the tool that’s already in front of them. They don’t switch to a different chatbot to do procurement.

So the platform winning the workday is also winning the B2B research session. Consumer-weighted charts can’t see that, because consumers don’t spend their day in an enterprise coding assistant.

Why a ChatGPT-first strategy is now a liability

For two years, “optimize for AI search” has quietly meant “optimize for ChatGPT,” and for a while that was defensible. At 89% of B2B referrals, ChatGPT was the channel.

At 63% and falling, it isn’t anymore. A ChatGPT-only program now ignores more than a third of where B2B buyers actually arrive from, and the ignored share is the part that’s growing. That’s not a gap you can wait out. It’s a gap that widens every month you treat the other engines as afterthoughts.

The problem compounds because the engines don’t read the same web. Optimizing for ChatGPT does not passively earn you Claude visibility. Claude pulls its results from Brave Search, not Bing or Google, and its citations barely overlap with what ChatGPT surfaces. The pages that win on one engine routinely lose on the next, which is why we keep arguing that you have to report each engine on its own line, never blended into a single score. The fragmentation in traffic and the fragmentation in citations are the same story told from two ends: there is no longer one AI surface to win.

This sits on top of a measurement problem that’s already embarrassing. A 2026 survey found only 14% of marketers track AI search citations at all, even as 89% of brands already appear in them and 43% call AI search a core strategy (GlobeNewswire). Most teams aren’t tracking ChatGPT carefully, never mind Claude. The fragmentation is making a blind spot they already had much larger.

What to do about it

This isn’t a call to chase every chatbot equally. It’s a call to weight your effort to where your buyers are, which for B2B is no longer one place.

1. Re-weight by B2B reality, not the blended chart. Stop sizing platforms by their share of all web traffic. Size them by share of your category’s AI referrals. If you’re B2B, that almost certainly means Claude deserves real attention and Gemini deserves more than you’re giving it.

2. Track Claude as a first-class platform today. If your AI visibility reporting covers ChatGPT and Google and stops there, you’re blind to your fastest-growing B2B referral source. Add Claude to the dashboard before it’s 25% of the channel instead of 18%.

3. Optimize for each engine’s retrieval, separately. Claude leans on Brave Search; ChatGPT rewrites prompts and leans encyclopedic; Perplexity stays close to the literal query and favors forums. One content plan can’t serve all four. Earn the source types each engine actually pulls from.

4. Watch the slope, not just the level. The headline finding here isn’t “Claude is at 18%.” It’s that Claude went from 1% to 18% in eight months. Static snapshots miss the move. Track the trend per platform so you see a channel rising while it’s still cheap to win.

5. Instrument referrals, not just citations. Citations tell you who AI quotes. Referral traffic tells you who AI actually sends to your site. Tag claude.ai, perplexity.ai, gemini.google.com, and chatgpt.com as distinct sources in your analytics so the shift shows up in your own data, not just someone’s industry report.

The takeaway

The AI search market isn’t consolidating around one or two winners. In the segment that matters most to B2B brands, it’s doing the opposite — fragmenting fast, with a surprise runner-up that the consumer-weighted charts are built to overlook. ChatGPT is still the largest single source. It is no longer the only one that counts, and the trend line says that gap keeps closing.

The brands that win the next year of B2B AI search will be the ones who noticed that “optimize for AI” stopped meaning “optimize for ChatGPT” sometime around the start of 2026, and re-pointed their effort, and their tracking, at the four engines their buyers actually use.

RivalHound tracks your brand’s visibility across ChatGPT, Google AI, Perplexity, Claude, and more — each engine reported on its own line. Start monitoring to see where you stand.