Rankings Up, Traffic Down: What the AI Search Paradox Means for Your Business

Your rankings are up. Your traffic is down. Your SEO agency says everything's fine.

Welcome to the AI search paradox.

If you own or run marketing for a business that's been investing in SEO for a few years, there's a good chance you've seen this pattern in your own Search Console: the positions holding or improving while the clicks quietly slide. It's not your imagination, it's not a penalty, and it's not your agency slacking off. The search results page itself has changed underneath you.

The data backs you up

This isn't anecdote. Across the industry, the studies all point the same way:

• Position 1 in Google used to be worth around a 7.6% click-through rate. When an AI Overview sits above it, that drops by more than half — down to roughly 1.6% (Ahrefs, Feb 2026).
• In Google's AI Mode, initial studies put the zero-click rate above 90%.
• And here's the number that should worry every business owner: roughly 91% of brands winning traditional Google SEO currently have zero visibility in AI search results.

That last one is the paradox in a single sentence. You can be brilliant at SEO — rankings, links, technical foundations, the lot — and still be invisible where a growing share of your customers' questions are actually being answered.

What we're seeing hands-on

I run an SEO and AI search agency, and the clearest example I've seen is in the property sector. Search for "best estate agent in [town]" and the results are dominated by an AI Overview. But here's what most people miss: the sources the AI Overview cites and the websites actually ranking in the traditional results are often vastly different lists. A business can be absent from page one and still be cited by the AI answer — or sit at number one organically and never get a mention.

That matters because the AI Overview dominates the same surface as the rankings. It's not a side panel you can ignore while the "real" results do their work underneath — on commercial queries like these, the AI answer is the results page. The battleground hasn't moved to a new field. It's the same field, and the rules of engagement changed mid-match.

The mental model mistake most businesses make

Most clients are now used to AI Overviews appearing in search results — that's not the hard part. The difficulty is creating a separate channel in your head that treats AI search as genuinely different from "the ten blue links."

Every previous SERP change — featured snippets, local packs, shopping carousels — was absorbed into what we already called SEO. AI is different, and it needs to be treated differently. But here's where I'll push back on the industry's current direction: treating AI search optimisation (call it AEO or GEO) as a completely separate specialty is dangerous too.

I think of it as SEO+. The foundations — technical health, authority, genuinely useful content, structured data — carry over more than most commentary admits. AI search rewards the same things traditional search rewards, plus a layer on top: consistent brand information across the web, citations and reviews on third-party surfaces, and content structured so an AI engine can extract and trust the answer.

Do the foundations badly and you'll lose both games. Ignore the plus layer and you'll win rankings nobody sees.

"Should we still be spending on SEO if AI is taking the clicks?"

I get asked this weekly, and my honest answer is: yes — the tactics cross over a lot. A site with strong authority, clean structure and real answers to real customer questions is exactly the site AI engines favour citing. Cutting SEO spend because of AI would be like stopping maintaining your shopfront because people started asking their friends for recommendations.

But — and this is the part that can be genuinely deadly — do not assume SEO spend covers you for AI visibility. The evidence above cuts both ways: the crossover means SEO helps, not that SEO suffices. Brands are getting cited with no rankings at all, while other brands rank beautifully and get ignored by the AI answers sitting on top of them.

What to actually do about it

If you're spending on SEO in 2026, ask yourself (or your agency) three questions:

1. Where do we appear when the query is asked to an AI engine? Not "how do we rank" — where do ChatGPT, Gemini and Google's AI Overviews actually name us? Test it with your ten most commercially important queries. Today.
2. Do our rankings and our AI citations match? If they diverge — cited but not ranking, or ranking but not cited — that gap is your map of what to fix first.
3. Is our brand information consistent everywhere an AI engine might look? Reviews, directories, social profiles, your own site. AI engines resolve brands from the whole web, not just your domain. Conflicting information about who you are and what you do is one of the biggest AI search risks there is.

The businesses that get hurt worst by the next two years of search change won't be the ones that did SEO badly. They'll be the ones that did reasonable SEO and assumed the world owed them traffic for it.

The paradox has a solution. Rankings still matter. Traffic is changing. And the winners are the ones treating AI search as SEO+ — not a new religion, not a fad to ignore, but the same discipline with a bigger board to play on.

Tom Marriott runs Odd Duck Digital, an AI-first SEO and digital marketing agency. Odd Duck helps growing businesses win both the rankings and the AI answers.