July 10, 2026: The Day Blue Links Stopped Being the Default

The June post on this site, The Death of the Blue Link, made a specific claim: blue links were not dead, they were demoted. That was the correct call at the time. It was also a hedge, because “demoted” is a direction, not a line you can point to on a calendar.

Now there’s a line.

On July 10, 2026, Google finished rolling out AI-generated answers as the default primary output for search. Not a test bucket. Not a feature that shows up on some queries and not others depending on Google’s mood that week. Default. The written-by-Gemini answer sits at the top, the ranked list of ten links sits under it, and on a meaningful share of queries the list doesn’t make it onto the first screen at all.

That’s the fact. What follows is what it actually changed, and what it didn’t.

The number that moved first

AI Overviews went from appearing in roughly 15% of Google searches to 43%, over the twelve months leading into Similarweb’s 2026 Generative AI Landscape report, which TechCrunch covered on July 27. That’s not a July 10 number specifically. It’s the runway that made July 10 possible: adoption had already climbed high enough that flipping the default was closing a gap, not opening a new one.

Fifteen to forty-three in a year is the kind of growth curve that gets ignored while it’s happening and looks obvious in retrospect. It was happening the whole time this site was writing about “demotion.”

What that did to clicks

Two numbers, from two different measurements, both bad for anyone counting on organic traffic the old way.

Organic click-through on queries where an AI Overview is present dropped from 1.76% to 0.61%. That’s a 61% decline, and it’s specifically on the queries where the feature shows up, not an average across all search.

Separately: users click through to a traditional blue-link result about 8% of the time when an AI Overview is showing. Without one, that number is 15%. Almost half the clicks, gone, on the exact query types where a business used to count on organic traffic showing up reliably.

Neither number is new information if you’ve been paying attention since spring. What July 10 did was remove the “but this is still only some searches” caveat. It’s not some searches anymore. It’s the default state of the product.

The part that complicates the clean story

Here’s what doesn’t fit the “blue links are dead” headline, and it’s worth sitting with because it changes the actual strategy.

A Search Engine Land analysis found that 43.42% of AI Overview answers contain links back to Google’s own search results, typically four to six of them. The AI answer isn’t replacing the blue link. In a huge share of cases, it’s citing the blue link, inline, as the thing that makes the answer credible.

That’s a different problem than extinction. If your page is one of the four to six links an AI Overview cites, you didn’t lose the click, you got compressed into a footnote position that some fraction of users will still expand and follow. If you’re not one of the four to six, you were never going to win the old-style ranking race for that query either. The AI Overview isn’t the enemy of ranking well. It’s a stricter filter on what “ranking well” has to mean now.

This is the same argument made in information gain optimization: the sources that get cited are the ones offering something specific enough to extract, not the ones that ranked adequately under the old rules.

The model keeps changing under the policy

One more thing worth tracking, because almost nobody covering July 10 mentioned it: the July 10 default was written by Gemini 3.5 Flash. It has not stayed that model.

Google shipped Gemini 3.6 Flash on July 21, eleven days later. It rolled Gemini 3.7 Flash into the AI Mode model picker for Pro and Ultra subscribers on August 14. By early September, AI Mode had moved again, to 3.8 Flash.

Four models in under two months, powering the same “default AI answer” policy that hasn’t itself changed since July 10. The structural shift and the technical implementation are two separate things moving at two separate speeds. The policy is settled. The model writing your competitors’ summaries this month is not the same one that will be writing them next month, and whatever citation or extraction behavior you’re optimizing against today has a real chance of shifting slightly with the next swap.

Don’t build a strategy around the specific quirks of Gemini 3.8 Flash. Build one around being the kind of source any version of this system wants to cite.

What actually changes starting now

If you were waiting to see whether this was a permanent shift before committing resources to it, July 10 is the answer. It’s permanent enough to plan around.

That doesn’t mean abandon what worked before. Genuinely useful commercial and transactional pages, the ones where a user has already decided to buy and is comparing options, are still landing clicks at close to the old rate. This shift hit informational queries hardest, the “what is,” “how do I,” “why does” category, exactly where an AI answer with citations satisfies the intent completely.

What it means concretely:

Stop measuring success by ranking position alone on informational content. A page ranking first for a query an AI Overview fully answers is a page that isn’t getting clicked regardless of position.

Start checking whether your pages show up in the four to six citation slots inside the AI Overview itself, not just the traditional results below it. That’s a different, narrower competition than classic SEO, and most sites aren’t tracking it yet.

Keep publishing content built for citation over content built for ranking. Specific numbers, direct answers stated plainly near the top, claims a system can verify against something concrete on the page. See how to optimize for Google AI Overviews and the broader framework in AI Overviews SEO strategy for the specifics.

The June post got the direction right. July 10 is just the date the direction stopped being debatable.