The Biggest SEO Trend Everyone Got Wrong in 2025

Every year has a prediction that gets recycled at every conference until it sounds like fact. For 2025, that prediction was simple and confident: AI search would replace Google this year. ChatGPT and Perplexity would eat Google’s lunch, organic traffic would collapse, and SEO as we knew it would be over.

It did not happen. I find this one a little funny and a little sad, because a lot of smart people repeated it and a lot of budgets moved on the strength of it. Let me walk through what the data actually showed.

The prediction that aged the worst

The pitch was everywhere, in slide decks with a hockey-stick chart and keynote speakers telling rooms full of marketers to abandon Google and chase the chatbots. The framing was always the same: the old channel is dying, the new one is taking over, so move now or get left behind.

I am not anti-AI search; it is real and growing and it matters. The problem was the size of the claim. “Replace Google” is a specific, measurable thing, so I went and checked the numbers, the same way I did with my llms.txt research when the adoption claims did not smell right. The gap between the hype and the data was enormous.

What the data actually showed

Start with referral traffic. Per Similarweb’s clickstream data, Google still sends the large majority of all referral traffic across the web, and every AI assistant combined is only a low-single-digit share¹. ChatGPT alone, the biggest of them, is a fraction of a percent of search referrals.

On market share the story is the same. Google still holds around 90% of search. After a year of “Google is finished” headlines, the number barely moved.

So no, AI search did not replace Google in 2025, and it did not come close. If you reallocated your entire strategy off Google because of a conference talk, you spent the year chasing a sliver of traffic and ignoring the channel that still sends the vast majority of it.

None of that makes AI search fake. It just means the marketing oversold a real but much smaller story.

The real story nobody put on a slide

The genuinely interesting part is the one almost nobody predicted. Inside the slice of traffic that does come from AI assistants, the ranking got scrambled.

ChatGPT leads that niche at about 62.6% of AI-assistant referrals², which is roughly what you would expect. The surprise was second place: Claude jumped to about 18.5%, ahead of Gemini at about 10.6% and Perplexity at about 7.3%. Perplexity was the name everyone shouted in 2024, and in the actual data it came in fourth.

That reshuffle is the kind of thing the “AI will replace Google” crowd missed entirely, because they were busy making a giant prediction instead of watching the small market that actually existed.

The volume also hides something about quality. AI-referred traffic is tiny, but it converts much better than average, with ChatGPT referrals running around 14 to 16% in the data I saw. Someone who asks a chatbot a question and then clicks through is often further down the funnel than a casual searcher.

So the accurate 2025 takeaway was never that Google is dead. It was that a small, high-intent channel is forming, the leaderboard inside it is not what you expected, and you should be tracking it. That is genuinely useful to know, and it is also a lot less exciting than a hockey-stick slide, which is most of why it did not spread.

Why the hype spread anyway

Two things kept it going. The first is that “X is dead” travels a lot faster than “X is mostly fine but keep an eye on this niche,” because fear sells tickets and gets reposted. The second is that there was money in it: vendors had AI SEO tools to sell, and “Google is over” was a great way to sell them.

Which leads to the part nobody likes to talk about.

The hangover: overspending on AI SEO tools

A lot of enterprises overspent on AI SEO tooling across 2024 and 2025. The takeover narrative made it feel urgent, so budgets got approved fast. Then the ROI came in below the pitch, because the channel those tools were chasing was a few percent of traffic, not a replacement for the whole funnel.

By 2026, buyers got more skeptical, which is healthy. The skepticism is not really about AI so much as about being overpromised. People got burned once, and now they want to see the number before they sign the contract.

I dig into which of those 2026 promises I am personally skipping in the SEO trends I am ignoring in 2026. I do not buy a tool just because a keynote scared me.

What to actually do in 2026

The lesson is not to ignore AI search, which would just be the same mistake in reverse. The lesson is to size it correctly.

• Plan for AI search as an emerging, high-intent channel. It converts well and it is growing, so track your referrals from ChatGPT, Claude, Gemini, and Perplexity and watch the trend line.
• Do not bet the company on a takeover that is not in the numbers. Google still moves 88% of referral traffic, and your core SEO work still pays the bills.
• Treat every “X is dead” headline as a prompt to go check the data yourself, since that is most of the job. If you cannot find the number, the headline is probably selling something.

If you want the practical split between optimizing for AI Overviews and doing classic search work, I covered that in AI Overviews vs classic SEO. And for where I think the real 2026 movement is, the pillar is SEO Trends 2026.

The takeaway

The biggest 2025 trend everyone got wrong was a confidence problem, not a technology problem. AI search is here. It is just a high-intent niche, not a Google-killer, and the data said so the entire time. Anyone who measured instead of repeating the slide had a quietly good year.

Measure first, then decide. That is the boring advice that keeps working.

Frequently asked questions

Did AI search replace Google in 2025?

No. Google still sent the large majority of all web referral traffic in 2025 per Similarweb, and held around 90% of search market share. Every AI assistant combined was a low single-digit share of referrals. The replacement narrative was repeated at conferences, but the data never supported it.

Is ChatGPT bigger than Google for search?

Not close. ChatGPT is only a fraction of a percent of search referrals, while Google holds around 90% of search and the large majority of referral traffic. ChatGPT is the largest AI assistant by referrals, but that is a small slice of a small channel, not a rival to Google's overall scale.

Are AI SEO tools worth it?

Sometimes, but check the numbers first. Many enterprises overspent on AI SEO tools across 2024 and 2025 chasing a takeover that never arrived, and the ROI disappointed. By 2026 buyers got more skeptical. Buy tools that match the channel's real size, which is a few percent of traffic, not a replacement.

Which AI assistant sends the most referral traffic?

ChatGPT leads AI-assistant referrals at about 62.6%. The surprise was second place: Claude jumped to roughly 18.5%, ahead of Gemini at about 10.6% and Perplexity at about 7.3%. The Perplexity hype from 2024 did not match where the traffic actually landed.

Does AI search traffic convert well?

Yes, that is its real value. AI-referred traffic is small in volume but high in intent. ChatGPT referrals were converting around 14 to 16% in 2025 data, well above typical search. Treat AI search as a high-intent niche channel worth tracking, not as a replacement for your core Google traffic.

Sources

1. AI assistants remain a low-single-digit share of referral traffic, a small fraction of Google’s volume in clickstream data. Similarweb, Gen AI stats 2026. ↩

2. Brand-averaged share of measurable B2B AI referrals, March to April 2026: ChatGPT 62.6%, Claude 18.5%, Gemini 10.6%, Perplexity 7.3%. Goodie, 2026 AI Search Traffic Report. ↩