What is happening with Google AI Overviews and Generative search?

Google has shifted from listing results to summarizing them. AI Overviews and AI Mode in Search now generate answer-style summaries at the top of many results pages, pulling from multiple sources and Google’s own knowledge graph.

For publishers, brands, and agencies, that means the old model of “rank in the top 3 and you are fine” is no longer reliable. When an AI summary appears, users are roughly half as likely to click a traditional result as they are on a classic page without a summary.

This article explains what Google AI Overviews actually are, how they change click behavior, how they differ from tools like ChatGPT and Perplexity, and what to adjust in your SEO and content program so your site is still visible and cited when AI systems do the talking.

Key takeaways

AI Overviews and AI Mode generate synthesized answers from multiple web pages and Google’s knowledge graph, then show a few supporting links.
Click rates on traditional organic results are significantly lower when an AI summary appears, which directly affects traffic forecasts and reporting.
Google uses techniques like query fan out, where a single query is expanded into many related sub-queries, then recombines results into one answer.
Ranking well is necessary but no longer sufficient. Your content also has to be structured so it is easy to extract, quote, and attribute.
Agencies and in-house teams need a hybrid strategy that covers classic SEO fundamentals and a newer layer focused on AI visibility, citations, and answer quality.

What are Google AI Overviews and AI Mode?

AI Overviews are AI generated snapshots that appear above or within the search results when Google believes a synthesized answer would be especially helpful. They provide a short explanation with links so users can explore further on the web.

In May 2024 Google announced that AI Overviews would roll out to all users in the United States, with expansion to over a billion people globally. They have since expanded to many countries and languages and now appear on a wide range of informational and decision oriented queries.

AI Mode is the more AI heavy experience that removes most of the classic list of links and replaces it with a full AI answer panel and supporting citations. Under the hood, Google has confirmed that AI Mode uses a query fan out technique, where your question is broken into subtopics and many related queries are issued in parallel to build a richer answer.

In both cases your content is not displayed as a simple ranking position. It becomes one of several sources used to justify whatever the model decides to say.

How AI Overviews change click behavior and traffic

Pew Research analyzed real browsing data and found that when users see an AI summary on a Google results page, they click on a traditional search result only about 8 percent of the time, compared with roughly 15 percent when no AI summary is present. Source: Pew Research Center

Other analyses of that same data show that clicks on the cited links inside the AI summary are even rarer, often around 1 percent of visits. Users are more likely to read the answer, accept it, and move on without visiting the underlying sites.

For you, that means:

Forecasts based on historical click through rates will be too optimistic in categories where AI Overviews now appear frequently.
Some keywords may still show you as “visible” in traditional rank trackers, but your share of actual clicks and influence is much smaller than before.
Branded and comparison queries are not immune. AI Overviews now appear on a growing share of navigational and “brand vs competitor” searches.

This is not a temporary experiment anymore. It is a new baseline for how search works.

How AI Overviews differ from ChatGPT, Perplexity, and other AI search tools

It is tempting to lump everything together as “AI search,” but the underlying mechanics and incentives differ.

Google AI Overviews and AI Mode run on top of the existing search index, ranking systems, and ads infrastructure. They have to balance user satisfaction, ad revenue, and legal scrutiny while working inside the Google ecosystem.
ChatGPT, Perplexity, and other assistants often lean more heavily on query fan out, pulling from a wider mix of sources and sometimes showing more citations on screen, but they are still selective about what they display.
Perplexity distinctly markets itself as “answer plus citations,” while Google has historically emphasized speed, a clean interface, and a smaller set of outbound links.

For a brand, the pattern is consistent: whichever system the user chooses, a small number of sources shape the answer. If you are missing from that small set, you are missing from the conversation.

Why this matters for your SEO and content program

Traditional SEO fundamentals still matter. Google’s own documentation on AI features states that there is no special markup or separate optimization required beyond creating helpful, well structured, crawlable content that follows normal best practices.

The problem is that “good enough for classic SEO” is often not good enough to be selected as the justification inside an AI answer. AI systems favor:

Clear, answer first sections that directly address a specific question.
Well defined entities such as products, integrations, audiences, and constraints.
Bounded claims with evidence, dates, and sources rather than vague marketing language.
Internal linking that makes it obvious which page is the ground truth for each topic.

In parallel, AI models are increasingly trained and evaluated against third party sources such as review sites, forums, Wikipedia style pages, and major publishers. If those surfaces misrepresent your brand or your competitors dominate them, AI summaries will reflect that.

What you can do now

You cannot control when an AI Overview appears, but you can control how easy it is for AI systems to quote you accurately. A practical starting point:

Identify your high value queries

Start with the handful of query themes that matter most to revenue and pipeline, such as:

Category plus “best for” and “alternatives.”
Your brand plus “pricing,” “reviews,” “vs competitor,” and “integration.”
Implementation, security, and compliance questions that show strong purchase intent.

Test these in Google and in AI assistants to see whether AI summaries appear, which brands are mentioned, and which pages are being cited.

Make key pages answer first and quote ready

Focus on the small set of pages that should represent you in those high value answers:

Category and solution pages.
Product and integration pages.
Comparison and “best for” pages.
Pricing model and implementation overview pages.

Restructure them so that each major question is clearly answered near the top, with constraints, examples, and internal links to deeper documentation. This reduces ambiguity when AI systems extract snippets.

Strengthen authority and third party corroboration

Improve how you are represented on the external surfaces AI relies on:

Review and comparison sites in your category.
Partner listings and marketplaces.
High quality PR, case studies, and expert commentary on reputable domains.

AI systems are more likely to trust your claims when they see consistent facts repeated in multiple credible places.

Monitor AI Overviews and AI chat outputs

Build a small, repeatable prompt set and track it over time. For each query, record:

Whether you are mentioned.
Whether you are cited or linked, and to which pages.
How you are positioned relative to competitors.
Any factual errors or risky claims.

Use that data to decide which pages to upgrade next and where you need better documentation or third party validation.

How Potenture approaches AI search visibility

At Potenture we treat generative experiences as an additional layer on top of classic SEO, not a replacement for it. Our work typically focuses on four pillars:

Structure for extractability: answer first content, prompt based headings, comparison ready sections, and micro guides around narrow buyer questions.
Entity clarity: consistent naming for your brand, products, modules, integrations, audiences, and constraints so AI systems can attribute correctly.
Evidence and trust: disciplined claims, proof blocks, citations, medically or technically reviewed content where needed, and freshness signals.
Technical and architectural clarity: crawlable internal linking, stable canonical pages for each intent, and schema where it reinforces entities rather than gaming rich results.

From there we baseline your current AI visibility, identify the initial set of pages and third party surfaces to fix, and ship changes in short sprints instead of attempting a full site rewrite.

AI Overviews and generative search are not going away. The brands that adapt fastest will not just keep some of their historical traffic. They will become the voices AI systems rely on when explaining the category to everyone else.