Where the Traffic Went

64 publishers. Organic clicks down 42% since AI Overviews scaled. Breaking news and Discover up.

The chart did not look like a bad month.

It looked like a contract ending.

For years, the deal was simple. Publishers, companies, and creators gave Google the web’s raw material. Articles. Reviews. Research. Comparisons. Definitions. Tutorials.

Google organized it, ranked it, and sent some of the attention back.

The deal was never equal. But it was legible.

That contract is changing.

AI search does not need to send the user through the same door. It can read the page, take the useful part, summarize the answer, and place the source beside the claim.

Sometimes the citation is visible. Sometimes the brand gets named. Sometimes the page helps build the answer without becoming the destination.

The traffic did not disappear.

It moved.

Some moved into the answer. Some moved into citations. Some moved to channels Google still controls, or to brands people already trust enough to visit directly.

The old question was, “Why did our traffic fall?”

The better question is:

Where did the value go?

This is the part many traffic reports still hide.

They show the loss. They do not show the exchange rate.

A click was simple to count. A visit became a session. A session became a chart. A chart became confidence.

But AI search breaks that measurement comfort.

A page can lose the click and still shape the answer. A company can lose traffic and gain visibility. A publisher can be cited and still lose the business model that paid for the reporting.

That is the uncomfortable part.

The new contract does not pay everyone equally.

It rewards sources strong enough to be selected, brands strong enough to be remembered, and publishers fast enough to own moments before AI can flatten them into summaries.

Everyone else gets compressed.

Right Answers, Thin Trust traced this shift from the reader’s side. This is the company’s.

The new currency is recognition.

Not the kind people give. The kind an AI system has already learned to associate with you before the question is asked.

When an AI answer is assembled, it does not begin with your latest page alone. It draws from the patterns already attached to a topic. Those patterns build slowly, through pages, mentions, citations, links, authors, interviews, profiles, and the quiet repetition of being named beside the right ideas.

The brand that appears in the answer is rarely the one that wrote the best page that week.

It is the one the system already understands.

This is the part most GEO advice misses.

You cannot retrofit recognition. You cannot add it at the end with schema, rewritten FAQs, or a page that sounds like your competitors. By the time the answer is generated, the system has already formed a rough map of the category.

A new player has to teach that map from zero.

Three groups are pulling away.

First, sources with original data. They are where facts begin.

Second, brands that were already trusted before AI search arrived. They came pre-loaded.

Third, authors with a named identity across platforms. The system can attach the claim to a person it has seen before.

Everyone else is asking a stranger to vouch for them.

There is a deeper change inside the change.

The old web was a system of access. If your page could be reached, it could be ranked. Effort produced visibility. A new site with the right answer could climb in months.

The path was hard. But the path existed.

The new web is a system of attestation. The answer is built from sources the system has been taught to trust. Trust accrues slowly and unevenly. It tends to settle on names that were already settled.

Access was open by design. Attestation is closed by default.

This is why the middle is in the most pain.

The largest publishers will keep being cited. Their names are already attached to the category. Their pages are everywhere. Their authors are visible across the web. They are the safe choice for the system.

They get safer.

The smallest creators can route around it. A loyal audience. A podcast. A community. A newsletter. They were never fully dependent on Google sending strangers through the front door.

The middle had a strategy.

The strategy was the open web.

That strategy is now harder to defend.

This is not nostalgia. It is a description.

The next ten years will be kinder to those who already had institutional trust and those who own a direct audience.

Everyone in between has to choose.

The choice is not a tactic.

It is an identity question the open web let companies avoid.

For two decades, the open web allowed a middle position to exist. You could be a brand without becoming a category name. You could publish without being a publisher. You could rank without earning a seat in the market’s memory.

Google’s distribution kept that middle alive.

The middle was never a strategy. It was a side effect of cheap distribution.

That side effect is ending.

The two real positions left are demanding. Each asks for a different muscle, a different patience, and a different way to measure the year.

Institutional trust is built over years through citations, mentions, primary research, named expertise, and the slow accumulation of being correct in public. It is expensive. Most of the cost is paid before the return appears. The companies that own it now bought it ten years ago, often without knowing what they were buying.

A direct audience is built through repeated permission. Email lists. Communities. Podcasts. Events. Products people return to without searching.

The cost is also long. But it compounds differently.

Not visibility to strangers. Repetition with the same people.

Both work.

Neither is the strategy most companies in the middle are running today.

Most are still publishing as if Google will send the audience back.

It will not send it back in the same numbers, for the same queries, or on the same terms.

The honest move is to look at what you are actually building and decide.

The 42 percent was not the story.

The story was the contract behind it.

Google built a large part of the web’s commercial distribution on a deal.

You make useful content. We send useful attention.

Both sides got something. Both sides understood the terms.

That deal is being rewritten in public. Not announced. Not negotiated. Just edited, query by query, until the old version no longer applies.

The companies asking why their traffic fell are reading the wrong document.

The traffic moved because the terms moved.

The new terms reward two kinds of businesses: sources the system has learned to trust, and companies with audiences they can reach directly.

Two positions. Two strategies. Two ways to build without assuming someone else will deliver the audience.

Everything else is the residue of a contract that no longer exists.

Decide which one you are building.

Then build it for ten years.

That is the only honest answer to “where did the traffic go.”

Source: the 42% click decline is from Define Media Group’s analysis of 64 publisher sites, reported by Search Engine Land.

I write about search, AI, and the questions worth sitting with.