Google now answers the question. What happens to your site?

Somewhere between 2024 and now, the deal that built the open web quietly changed. You wrote something useful, Google sent you readers, and the readers occasionally became customers. That trade is coming apart.

The numbers are not subtle. Pew Research watched 68,879 real Google searches and found that when an AI Overview appears, users click a traditional result 8% of the time — against 15% when there's no overview. A randomised field experiment put the drop in outbound organic clicks at 39.8%. SparkToro's 2026 analysis found that fewer than a third of Google searches now end in a click to the open web at all.

Whatever the exact figure, the direction is the same everywhere you look, and AI Overviews now trigger on roughly half of all queries.

What this actually means for a small business

The panic reading is "SEO is dead." That's wrong, and it's an expensive thing to believe.

Here's what's really happening. Search is splitting into two jobs. The first is the one you know: someone types a query, scans results, clicks. That job is shrinking. The second is new: an assistant — Google's AI Overview, ChatGPT, Claude, Perplexity — reads the web on the user's behalf and synthesises an answer. Your business either appears inside that answer, correctly described, or it doesn't exist in the conversation.

The second job is the one almost nobody is optimising for.

A ranking gets you a click. Being quotable gets you recommended. They are not the same skill.

What being quotable looks like

When we rebuilt ChimpFit, the SEO work was table stakes. The part that mattered more was making the site legible to a machine that has to summarise it in two sentences.

That means, concretely:

• Answer the actual question, in full, on the page. Not a teaser that funnels to a contact form. An assistant cannot quote a call-to-action.
• Publish the specifics you'd normally hide. Prices, timelines, what you don't do. We publish every price we charge partly for this reason — a model can cite "$1,500 for a five-page site" and be useful. It cannot do anything with "contact us for a quote."
• Structured data that matches the visible page. Schema.org markup describing the same facts a human reads.
• A machine-readable summary of the business. We keep an answers page and an llms.txt for exactly this.

None of that is a trick. It's the opposite of a trick — it's being specific in public, where vagueness used to be safe.

The uncomfortable part

This is worse for some businesses than others. If your traffic came from broad informational content — "what is a roof underlayment" — a large share of it is not coming back, because that's precisely the query an AI Overview answers completely.

If your traffic came from people with a problem and a budget, in a specific place, you're in better shape than the headlines suggest. Assistants still hand off high-intent, local, transactional queries. They just hand off fewer, better ones.

So the honest strategy isn't "make more content." It's make fewer pages that are dramatically more specific, and make sure every fact a machine might need is stated plainly rather than implied.

Where we'd start

If you're watching traffic fall and you're not sure how much is you and how much is the industry, that's a diagnosis worth doing before spending anything. Our marketing and AEO work starts with exactly that audit — where you stand in classic search, and separately, whether the major assistants can describe your business correctly at all. Usually the second answer is worse than the first, and it's cheaper to fix.

You can also just ask ChatGPT or Claude what your company does, and see whether the answer is one you'd have written. That test costs nothing and tends to be clarifying.

---

Sources: Pew Research via Search Engine Land · randomised field experiment on AI Overviews · SparkToro, 2026 click analysis · Amsive study on CTR decline