Search Generative Experience: What Happened

The search generative experience (SGE) was Google’s 2023 opt-in test of AI-written answers at the top of search results. It never launched under that name. Its core idea became AI Overviews in May 2024, and the conversational half grew into AI Mode in 2025.

This page is for marketers and owners who still see “SGE” in old advice and want to know what became of it. It covers the timeline, what the data says about clicks, and which SGE-era tips still hold up.

On this page

• The SGE timeline in four dates
• What was SGE at launch?
• How did SGE become AI Overviews?
• Where AI Mode came from
• Why does SGE still show up in search results?
• What happened to clicks
• Which SGE-era advice still holds and which is retired
• Where the search generative experience leaves you

The SGE timeline in four dates

Four Google announcements explain the whole arc. The diagram below puts them in order.

Scroll the diagram sideways to see all of it.

What was SGE at launch?

SGE was an experiment you had to sign up for. Google’s announcement on May 10, 2023 introduced it as part of Search Labs, available in Chrome desktop and the Google app in the U.S., in English only.

Two features defined it:

• The AI snapshot. A generated summary of key information to consider, with links to go deeper.
• Conversational follow-ups. Suggested next questions that carried context from one query to the next.

Google was candid that it would not always be right. The post said Search “will not always get it right” and that the company was limiting the types of queries where the features appeared. That restraint matters for the history: SGE was a controlled test, and the advice written about it was written about a product most searchers never saw.

How did SGE become AI Overviews?

SGE became AI Overviews when Google moved the feature out of Labs and into the main results page for all U.S. users. The Google I/O announcement on May 14, 2024 said AI Overviews would begin rolling out to everyone in the U.S. that day, with more countries soon after and a goal of reaching over a billion people by the end of the year.

The post describes AI Overviews as growing out of the Search Labs experiment, which people had used billions of times. It does not use the SGE name. The experiment label simply disappeared, which is why so much older content still says “SGE” for something Google now calls something else.

Google made two claims in that post that are worth keeping in mind, because they are claims, not measurements anyone outside Google can check:

• Links inside AI Overviews get more clicks than the same page would as a traditional listing for that query.
• People visit a greater diversity of websites when they ask complex questions.

We cover what independent studies found, and how to get included, in how to appear in Google AI Overviews. Here the point is narrower: the SGE name stopped mattering the day the product shipped to everyone.

Where AI Mode came from

AI Mode is the conversational half of SGE, rebuilt as its own destination. Google’s March 5, 2025 post introduced it as a Labs experiment that extends AI Overviews with more advanced reasoning and multimodal input, running on a custom version of Gemini 2.0. Google One AI Premium subscribers were the first invited.

Its defining technique is query fan-out. Instead of running your one query, the system sends several related searches at once across subtopics and data sources, then combines the results into a single response.

On May 20, 2025, Google began rolling AI Mode out to U.S. users with no Labs sign-up, and added Deep Search, which can issue hundreds of searches and produce a cited report. The same post said AI Overviews drove over a 10% increase in Google usage for the queries that show them, in the largest markets including the U.S. and India. That too is Google’s own figure.

Notice the pattern across the four dates. Each step took the same idea, a generated answer with links, and moved it closer to the default experience while adding more machinery behind it. SGE showed one snapshot. AI Overviews put summaries on the main page. AI Mode splits a question into subtopics and runs many searches. For a site owner, that last change matters most: the pages that get used are the ones that answer a specific sub-question well, not only the ones that rank for the head term.

The practical differences between the two surfaces, and how to be a source in the fan-out, are in Google AI Mode explained. For where this sits among all the other 2026 changes, see AI SEO in 2026: what changed.

Why does SGE still show up in search results?

The name lingers because the content written about it kept ranking. The live results for “search generative experience” are still dominated by explainers dated 2023 and 2024, and several of them only mention AI Overviews as a parenthetical rename. A reader landing on one of those pages gets a description of a Labs test, not of the product on their screen today.

That has a cost for decision-making. Advice tuned to a limited opt-in experiment, shown to a small group on a narrow set of queries, was often applied as if it described the whole of Google. Before acting on any SGE-era article, check its date against the four announcements above, and check its claims against Google’s current documentation. If the article cannot be reconciled with the documentation, trust the documentation.

What happened to clicks

Clicks fell on results pages that show a summary, according to the best independent browsing data available. The diagram below charts the Pew Research Center’s numbers.

Scroll the diagram sideways to see all of it.

Pew’s analysis tracked 900 U.S. adults. It found AI summaries on 18% of the unique searches studied. Users clicked a traditional result on 8% of visits to pages with a summary, against 15% on pages without one. They clicked a link inside the summary on 1% of visits. And 26% of visits to pages with a summary ended after the search page, versus 16% without.

Two honest caveats. The data covers March 2025, so it describes a different product mix than today. And Pew compares different kinds of searches, so a summary is not necessarily the only cause of the gap. Our roundup of the other studies, which disagree on size, is in AI Overviews and the zero-click problem.

Which SGE-era advice still holds and which is retired

The advice that still holds is the plain Search fundamentals. The advice that has been retired is anything that treated the AI answer as a separate game with its own technical requirements.

Google’s AI features documentation is the cleanest reference. It says a page must be indexed and eligible to be shown in Search with a snippet to be a supporting link. It states there are no additional requirements to appear in AI Overviews or AI Mode, and no other special optimizations necessary. It also says all existing SEO fundamentals continue to be worthwhile, and lists the usual suspects: let Google crawl, link internally, keep key content in text, support it with quality images and video, and keep Merchant Center and Business Profile data current.

Here is how common SGE-era tips sort against that page.

The diagram below shows the same sorting in one glance.

Scroll the diagram sideways to see all of it.

One nuance on the retired side. Google says you do not need schema or AI files for its features. That is not the same as saying schema is worthless: structured data still supports rich results, and other AI engines read the web differently. Our take on that is in structured data for AI search and what llms.txt does and does not do.

A quick audit of your old SGE plan

If your team wrote an “SGE strategy” in 2023 or 2024, rework it in four steps.

1. Rename it. Replace “SGE” with two lines: AI Overviews (a summary on top of results) and AI Mode (a separate conversational tab). They behave differently, and your reporting should too.
2. Delete the exotic tasks. Anything that exists only to please a Google AI feature, such as special markup or extra files, can go. Google says it is not required.
3. Keep the fundamentals and finish them. Crawlable, indexed, snippet-eligible pages with the answer stated early. Our guide on content chunking for AI retrieval covers how to structure sections so they stand alone.
4. Re-baseline your traffic. Treat clicks on summary-heavy queries as a separate segment. Impressions and rankings can hold while clicks fall.

The old SGE advice also had a blind spot: it assumed Google was the only generative surface. It is not. Other assistants fetch and cite on their own rules, which is why our generative engine optimization guide treats Google as one source among several.

Where the search generative experience leaves you

The search generative experience left behind three durable lessons. A Labs test can become the default page within a year. A feature’s name can change while the rules underneath it barely move. And the winning habit is the unglamorous one: be indexed, be clear, be the page that answers the question, then measure what that earns you.

If you want a second pair of eyes on which of your pages are eligible, which are being cited, and what your old SGE plan should become, our AI SEO service starts with exactly that review.