AI SERPs Changes the Click. Does it Change the Visit?

If AI answers reduce clicks, what replaces the click as the unit of organic value?

That sounds like a reporting question. It isn’t. It’s a strategy question pretending to be a metric problem.

Because “fewer clicks” can mean at least three different things at once. It can mean you’re losing traffic. It can mean you’re losing the kind of traffic you used to rely on. Or it can mean something subtler: people still visit, but they arrive in a different mental state.

When a user does click after an AI Overview, are they arriving with clearer intent and higher readiness? Are they arriving to verify what they already saw, skeptical and quick to bounce? Or are they arriving fragmented, with most of what they wanted already satisfied on the SERP?

Those three visitors behave like three different audiences. If you treat them as one, your conclusions will be wrong.

And the mistake will look reasonable in a dashboard.

The default assumption in SEO reporting is simple: fewer clicks equals worse SEO.

Sometimes that’s true. But it’s no longer safe to treat it as a law of physics.

Multiple datasets over the past year have shown click-through rates dropping on queries that trigger AI Overviews, even when visibility stays high. BrightEdge has described impressions rising while click-throughs decline, with roughly a 30% click-through reduction since May 2024 in their readouts. Ahrefs reported a similar directionally negative impact in their CTR research and later updates. More recently, a randomized field experiment reported a meaningful reduction in organic clicks when AI Overviews appear, alongside a jump in zero-click behavior.

Here’s the trap: if fewer clicks is the new normal for some query classes, then “down traffic” is not automatically a content failure. It might be a distribution change.

The alternative framing is uncomfortable but useful: fewer clicks might mean fewer low-intent visits, while high intent visits stay stable or even get more valuable. The counterpoint is just as real: AI can reframe the query and rewrite expectations. If the SERP teaches people a new way to think about the problem, they may want a different experience than the one your page was built to deliver.

Same ranking. Different job.

We’ve been saying “Google is answering more on the SERP” for years. AI Overviews make that statement operational.

AI summaries don’t just steal a snippet. They change how research happens. They satisfy basic informational intent without a visit. They compress comparison research into a shorter consideration window. And they often introduce “next questions” that nudge users into a new branch of the journey that may not include your site at all.

The result is that clicks may drop, while the economics of acquisition shift. You still pay the cost of producing content. The SERP captures more of the value of that content by satisfying the first question in the interface.

Free traffic was never free. It was subsidized by the friction of having to click.

In-house teams are already seeing a pattern that feels like a glitch: rankings hold, impressions hold or rise, clicks fall.

If you diagnose that as “we need to rewrite the page,” you might fix nothing and burn weeks. The system changed. The SERP is now doing more of the top-of-funnel work and, importantly, taking credit for satisfying the user.

This is the moment where “SEO is infrastructure” stops being a slogan and becomes a budgeting reality. If your reporting still treats rankings plus traffic as the whole story, you’ll be blind exactly when leadership needs clarity most.

Best-case scenario, you get fewer visits but better ones. Conversion rate rises. Sales efficiency improves. Support-heavy, low-intent users disappear. In that world, the “traffic drop” is a feature you should celebrate quietly.

Worst-case scenario, you get fewer visits and worse downstream performance. The remaining visitors are pre-informed, more selective, and harder to persuade. They skim. They compare you against a mental model the AI summary just gave them. They bounce faster because they’re not here to learn; they’re here to decide.

Those two worlds can look identical in Search Console if you only stare at clicks.

AI answers cite facts, not trust.

Even when your site is used as a source, the user experience is often “Google told me” not “this brand taught me.” Google’s AI features can link to sources, but the SERP’s primary value is the synthesized response, not your byline.

If users don’t need to learn from you, you have to give them a reason to choose you.

That reason is rarely another paragraph that answers the same question the SERP already answered.

Aggregated AI responses can concentrate attention on a smaller set of canonical sources. Plenty of pages will still rank and still technically exist in the result set, but they’ll function more like inventory than outcomes. Decorative rankings.

This is why “visibility” is becoming less predictive of “value,” which is the opposite of how most SEO orgs are staffed and evaluated.

Traffic decline can be a feature when the decline is mostly unqualified demand. Fewer tire-kickers. Better CAC. Cleaner pipelines.

Traffic decline can also be a warning when discovery and education get absorbed on-SERP and your site is left with scraps: the last step of the journey without the earlier steps that created preference.

The question to move your team toward is not “did clicks drop?” It’s “did visits that matter drop, and where did the intent go?”

Clicks are not dead. They’re just demoted. They’re now one step in a chain you need to instrument end-to-end.

Start by treating “value per visit” and “value per impression” as first-class metrics. If impressions rise while clicks fall, the impression becomes the scarce asset you can still win. The question becomes whether you can convert that exposure into outcomes, even with fewer sessions.

Then segment by query class. Informational and how-to queries behave differently under AI answers than comparison queries, navigational brand queries, or problem-aware queries. We’re already seeing industry research that the click impact is not uniform and tends to be harsher on non-branded informational patterns.

Inside the site, track intent quality signals that don’t require guesswork. Watch engagement depth in a way that maps to your funnel, not vanity time-on-page. Track how quickly users reach a key event. Track internal navigation that indicates “second question” behavior. Track return rate and assisted conversions, because “verification visits” can look like short sessions and still matter later.

Finally, build one SERP-era efficiency view that your leadership can understand without believing in SEO magic. Impressions flow to clicks, clicks flow to qualified sessions, qualified sessions flow to revenue or a proxy you trust. Then you monitor where the drop actually happens.

If impressions stay flat but clicks fall, it’s likely a SERP behavior shift. If clicks stay flat but qualified sessions fall, it’s likely an intent mismatch on your landing experience. If qualified sessions hold but revenue falls, your positioning and proof are the problem.

Different failures. Different fixes.

Most teams will try to solve this with more content. That’s understandable. It’s also often the wrong first move.

Content strategy has to shift from “answer the question” to “earn the next step.” If the SERP handles the basics, your page needs to justify the visit by enabling action, comparison, implementation, or trust-building faster than the user can get it elsewhere.

Information architecture matters more because visitors arrive pre-educated. They don’t want the intro. They want the deeper branch immediately. If your pages force them through the beginner path, they leave.

Trust signals stop being decorative. Authorship, sourcing, freshness, and proof matter because AI summaries flatten nuance. You need receipts, not vibes.

And technical SEO still gates everything. Crawlability, rendering, indexing reliability, canonicalization. None of that got easier because Google added AI. If anything, the margin for error shrinks when fewer clicks are available to waste.

AI Overviews shift who does the work and who captures the value.

Publishers and brands still invest in content. The SERP increasingly delivers the first-order value of that content directly. Meanwhile, the remaining clicks skew toward people closer to a decision, which can be good, but also raises the bar for your experience. The “free traffic” era was built on a user having to leave the SERP to get the answer. That tax is shrinking.

So your moat cannot be “being the answer.” The answer is now table stakes, and often the SERP provides it.

Your moat becomes being the best place to act after the answer.

Don’t panic about clicks. Diagnose intent shift.

Assume more of your visitors are pre-educated. Optimize for the second question, not the first.

Update reporting so rankings plus traffic is no longer presented as the full story. It’s a partial story that can be actively misleading in an AI-shaped SERP.

Then pick one high-impact query class and run a simple audit. Find where AI Overviews are present. Compare the before-and-after distribution of visits, not just totals. Look at whether conversion rate rises, whether assisted conversions change, whether time-to-key-event improves or collapses.

If you do that work, you’ll know which world you’re in. In one world, you’re getting fewer but better visits and you can reallocate effort accordingly. In the other world, the SERP is absorbing discovery and you need to rebuild how your content earns the next step.

Either way, you stop arguing about clicks and start managing outcomes.

Alignment: Ensuring AI behavior matches human goals values and safety constraints.

Learn More

One more thing: AI is only as good as it’s operator, and if you are reading this newsletter, you are better than most!

Till next time,

PS: Let me know what you think of this issue, or anything else here: [email protected]