Half of B2B software buyers now start their research inside an AI chatbot, not Google (G2, 2026). Most "AI is changing SEO" advice is either panic or stats recycled from the July 2025 peak that Google has since walked back. This is the version with the numbers attached: what AI search actually does to your traffic, what gets you cited, and a GEO checklist tied to evidence instead of vibes.

Key Takeaways - Google AI Overviews surged to roughly 25% of queries in July 2025, then Google pulled them back to under 16% by November (Semrush, 2025). - Pages cited inside an AI Overview get about 35% more organic clicks; pages not cited lose roughly 58% of their CTR (Seer Interactive / Ahrefs, 2025). - Across 75,000 brands, YouTube mentions show the single strongest correlation (0.737) with AI visibility (Ahrefs, 2025) , correlation, not causation. - 94% of B2B decision-makers used an LLM during their 2025 purchase (Forrester, 2025). AI traffic is small (about 1% of the web) but converts roughly 10x better than search.

How big is AI search actually? The honest numbers

AI search is smaller than the headlines and bigger than the skeptics admit. Google AI Overviews appeared in 6.5% of queries in January 2025, peaked just under 25% in July, then fell back below 16% by November as Google narrowed where it shows them (Semrush, 2025). AI platforms drove more than 1.1 billion referral visits in June 2025, up 357% year over year, with ChatGPT accounting for about 87% of that traffic (Similarweb, 2025). And yet AI platforms still account for only around 1% of total web traffic across major industries (Conductor / Similarweb, 2025–26). Meanwhile, 58.5% of US Google searches end without a click, and only 360 of every 1,000 reach the open web (SparkToro x Datos, 2024).

One thing to clear up, because it confuses everyone: you'll see "AI Overviews appear in 60% of searches" and "AI Overviews appear in 16% of queries" published in the same week. Both are real. The higher figures count an AIO showing up anywhere in a large tracked keyword set; the 16% figure comes from a panel measuring the share of actual queries. They're different metrics. Don't treat the gap as a contradiction, and be precise about which one you're quoting.

What AI Overviews do to your clicks

When an AI summary appears, people click a traditional result in about 8% of searches, versus 15% when there's no summary , roughly half (Pew Research, 2025). Ahrefs put the hit at around 58% lower CTR for the top-ranking page when an AI Overview is present (Ahrefs, 2025). Seer Interactive measured organic CTR on AIO queries falling 61% and paid CTR falling 68% (Seer Interactive, 2025).

Here's the part most posts skip. Pages cited inside the AI Overview get about 35% more organic clicks and 91% more paid clicks than pages that aren't (Seer Interactive, 2025). And roughly 38% of AIO citations come from a page already ranking in the organic top 10 for that query (Ahrefs, 2025). So the takeaway isn't "AI Overviews killed SEO." It's narrower and more useful: ranking still matters, and being one of the cited sources is the new position zero.

What actually gets you cited by AI engines

Ahrefs analyzed 75,000 brands and found that YouTube brand mentions show the strongest correlation with AI visibility: 0.737 across ChatGPT, AI Mode and AI Overviews, ahead of branded web mentions (0.66–0.71) and far ahead of how many pages your site has (about 0.19) (Ahrefs, 2025). Correlation isn't causation, and Ahrefs says so plainly. But the pattern holds across the data: AI engines lean on entities that get discussed across the web, not on sites that simply publish a lot of pages.

Engine behavior differs too. Perplexity cites roughly 22 sources per answer; ChatGPT cites about 8 (Qwairy, 2025). That means a single citation is worth more in ChatGPT than in Perplexity, and it changes how you'd prioritize. If you want the mechanics for each, we've broken down getting cited by ChatGPT in 90 days and the six factors behind Perplexity citations separately.

The B2B SaaS angle: you're being filtered out of the shortlist

For B2B SaaS, the AI shift isn't really a traffic problem. It's a discovery problem. 51% of B2B software buyers now start research with an AI chatbot, and 71% use one somewhere in the process, up from 60% the year before (G2, 2026). 94% of B2B decision-makers used at least one LLM during their 2025 purchase, and B2B buyers are adopting AI search at roughly three times the rate of consumers (Forrester, 2025). 69% of those G2-surveyed buyers picked a different vendor than they'd planned, based on a chatbot's recommendation.

Think about what that means. If the model doesn't surface you, you're not in the consideration set, and you'll never see the lost deal in your analytics. The flip side is encouraging: the traffic that does come through converts. LLM-referred visitors sign up at 1.66% versus 0.15% from organic search, and spend about 15 minutes on site versus roughly 8 from Google (Microsoft Clarity / Similarweb, 2025–26). Small, but high-intent. Worth treating seriously without panicking.

The GEO checklist: what to actually do

Generative engine optimization in 2026 is mostly the SEO you should already be doing, pointed at extractability and entity authority instead of just rankings. Tied to the data above:

• Get into the organic top 10 for the queries that matter to you. About 38% of AI Overview citations come from there (Ahrefs, 2025). If you want the distinctions, we've written up AEO vs GEO vs LLMO and why the difference matters.

• Build a YouTube presence. Mentions of your brand in video titles, descriptions and transcripts correlate most strongly with AI visibility (Ahrefs, 2025). One solid explainer per core topic, titled for the query, is a start.

• Earn branded mentions where models read. Reddit, Wikipedia, G2, credible industry roundups. AI engines weight entities discussed across third-party sources, not self-published volume.

• Structure content for extraction. Answer-first paragraphs of 40–60 words, one claim per sentence, FAQ blocks, clean headings. The same formatting that wins featured snippets is what gets you quoted by an LLM.

• Don't chase coverage you can't control. AI Overview presence is volatile and Google decides it. Optimize for being citable, not for triggering an AIO. For how AIOs behave in practice and how to respond, see how Google AI Overviews affect B2B traffic.

How to know if any of this is working

You can't manage AI visibility blind, and most teams are flying blind right now. Start with the cheap part: in GA4, build referral segments for chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. That's your AI referral traffic, and it's a real channel , more than a billion visits a month industry-wide (Similarweb, 2025). For citation tracking, the question of whether you actually appear in answers, tools like Profound, Ahrefs Brand Radar and Semrush's AI features sample prompts and report back. If you do nothing else this week, set up the GA4 segments. It takes ten minutes and you stop guessing.

A note on brand safety in AI-surfaced content

One thing pure-SEO advice skips, and it's where a security background helps: when an LLM surfaces content about you, it can also surface content about you that you didn't write. Prompt-injection-style manipulation of pages that AI systems read is a documented risk, not a hypothetical. If your category is contested, monitor what models say about your brand the way you'd monitor reviews. It's a security problem wearing a marketing hat, and it belongs on the same dashboard.

Frequently Asked Questions

Is SEO dead because of AI search?

No. About 38% of AI Overview citations come from pages already in the organic top 10 (Ahrefs, 2025), and pages cited inside an AI Overview get roughly 35% more clicks (Seer Interactive, 2025). Ranking didn't stop mattering. Being one of the cited sources became the new prize.

What's the difference between SEO, AEO and GEO?

SEO optimizes for ranking in the blue links. AEO (answer engine optimization) optimizes for being the answer in featured snippets and AI Overviews. GEO (generative engine optimization) optimizes for being cited by generative engines like ChatGPT and Perplexity. In practice they overlap heavily: answer-first, well-structured, authoritative content serves all three at once.

How much traffic comes from ChatGPT and Perplexity?

Industry-wide, AI platforms drove more than 1.1 billion referral visits in June 2025, up 357% year over year, with ChatGPT about 87% of it (Similarweb, 2025). For most sites that's still around 1% of total traffic, but those visitors convert roughly 10x better than organic search (Microsoft Clarity, 2025–26).

Do I need a separate strategy for AI search?

Mostly no. The fundamentals (earn authority, publish genuinely useful content, structure it for extraction, get into the organic top 10) cover SEO, AEO and GEO together. The genuinely AI-specific additions are building third-party brand mentions, especially on YouTube, and tracking AI referrals and citations so you know where you stand.

How often do Google AI Overviews appear?

It depends on how you measure. A panel of real queries put it under 16% by November 2025, down from a roughly 25% peak in July (Semrush, 2025). Larger keyword-tracking sets report higher numbers because they count an AIO appearing anywhere in the set. Either way, coverage is volatile and Google-controlled, so optimize for being citable rather than for triggering one.

The bottom line

AI search is real, small, growing, and high-intent. The playbook for it is about 80% the SEO fundamentals plus 20% genuinely new work: third-party brand presence, extractable structure, and actual measurement. The teams that fall behind are the ones still arguing about whether it matters instead of setting up the GA4 segments and getting their best pages into the top 10.