Which AI platform actually matters most right now?
I keep seeing businesses treat “ChatGPT, Perplexity, Gemini, Claude” as one undifferentiated blob to optimize for. It isn’t. The traffic behind those names is wildly uneven, it moved fast in the last year, and it should decide where you spend your limited AEO effort first.
By real traffic, in order: Google AI Overviews first (folded into Google Search itself, so it reaches the widest audience by far), then ChatGPT (still the dominant standalone assistant), then Gemini (rising fast on Google’s own distribution), then Claude (smaller, growing fastest), then Copilot and Perplexity, roughly tied, well behind the rest. That order should set your priority — not equal effort across six logos.
The order, with numbers
As of May–June 2026, across the AI platforms this site audits for clients:
| Platform | Share / traffic | Trend |
|---|---|---|
| Google AI Overviews | >40% of US Google searches (Similarweb) | sharply up year-over-year |
| ChatGPT | 53.9% of global AI-chatbot web visits | down from ~76% a year earlier |
| Gemini | 27.9% share, 2.9B monthly visits | up ~450% YoY, from 527.7M |
| Claude | 9.2% share, ~953M monthly visits | up ~855% YoY, from ~2.2% share |
| Copilot / Perplexity | well under 10% each | roughly flat, long tail |
Two things jump out. First, Google AI Overviews is not a fringe feature — it’s embedded in the search box nearly half of your buyers already use, which puts its real reach above every standalone assistant combined. Second, ChatGPT’s dominance is shrinking in real time. A year ago it was three-quarters of all chatbot traffic. Now it’s just over half, with Gemini and Claude eating the difference.
Same data, as a picture
Search infrastructure still dwarfs every standalone AI assistant combined — this is why AI Overviews can’t be treated as optional:
And among the standalone assistants people actually download or visit directly:
*AI Overviews has no standalone visit count since it’s embedded in Search — the 36B figure is an estimated reach (Google Search visits × estimated share of searches triggering an AI Overview), illustrative of scale, not a measured Similarweb metric. Everything else is measured monthly web visits, worldwide, per Similarweb-tracked data as reported in June–July 2026 coverage of its 2026 report.
The Google AI Overviews number is contested — say so
“Over 40%” is Similarweb’s own headline figure from its 2026 report, not a third-party estimate of Similarweb’s data. It is still not the only number out there. Adthena measured AI Overviews appearing in roughly 18% of US searches in June 2026, and an independent academic study put the activation rate at about 13.7%. Similarweb, Adthena, and academic researchers are measuring different things in different ways, and none of them publish a fully transparent methodology.
What all three agree on: the share is large, growing, and already reshapes how often a business gets seen without a click. What they don’t agree on is the exact number, so I’m giving you the range rather than picking whichever figure sounds most impressive.
What this means for where you spend effort
This isn’t an argument to ignore Claude, Copilot, or Perplexity — a citation on any of them is still worth having. It’s an argument against equaleffort across all six. If your content, structured data, and third-party signals aren’t at least solid on Google AI Overviews and ChatGPT before you worry about Perplexity, you are optimizing for the tail before the head.
Practically, that means: fix your schema and extractable content for how Google’s AI systems read your site first. Confirm ChatGPT can find and cite you second. Then work down the list — Gemini and Claude are worth real attention given how fast they’re growing, and Copilot and Perplexity last, not because they’re worthless, but because the volume behind them is still comparatively small.
What I changed on my own site because of this
I’d had the six platforms I audit listed as “ChatGPT, Perplexity, Google AI Overviews, Claude, Gemini, Copilot” across dozens of FAQ answers, titles, and meta descriptions on this site — an order that put Perplexity ahead of Google AI Overviews and Claude ahead of Gemini, both backwards from the actual traffic. That’s the same mistake I’m describing above, just made by me, on my own site, before I looked at the numbers carefully.
I reordered every instance to lead with Google AI Overviews, then ChatGPT, then Gemini, Claude, and the rest — same platforms, same claims, just honest about which ones carry the volume. Small fix, but it’s the same principle this whole site is built on: say what the numbers actually show, including when that means admitting your own content had it backwards.
What this doesn’t tell you
- Aggregate traffic isn’t your traffic. ChatGPT being 53.9% of chatbot visits doesn’t mean 53.9% of your potential customers ask ChatGPT about your category specifically. Check your own citation data per platform before assuming the global ranking applies to you.
- This is a snapshot, not a law. The ChatGPT-to-Gemini shift happened inside twelve months. A ranking this volatile will likely look different again by next year.
- Geography matters. The AI Overviews figure is specifically a US measurement; Canadian and other-market behavior may differ.
Questions about the method behind any of these figures, or want the sources listed individually? hami@hamitahm.com.
Want to know where you actually stand, platform by platform?
The audit checks Google AI Overviews, ChatGPT, Gemini, Claude, Copilot, and Perplexity individually — not a single blended score.
Run the free AI visibility check →