Google AI Overviews Are Changing Search: What Businesses Need to Know
There is a specific meeting happening in marketing departments all over the world right now, and it always goes the same way.
Someone pulls up the analytics. Impressions are flat or up. Rankings are stable, sometimes better than last year. Clicks are down thirty percent, forty percent, sometimes more. Nobody did anything wrong. Nothing broke.
Then somebody opens a private browser window, searches one of their top terms, and there it is. A generated summary occupying the entire first screen, answering the question completely, with a handful of links most people will never touch.
That is the whole story in one screenshot. Your ranking did not change. The value of your ranking did.
I want to be careful with this topic, because it attracts more panic and more bad advice than almost anything else in marketing right now. So let me tell you what actually changed, what Google actually says, and what I would do about it.
What AI Overviews Actually Are
AI Overviews are AI-generated summaries at the top of Google's results, built from multiple web sources with supporting links.
Three mechanics worth understanding.
They are grounded in Google's core ranking and quality systems rather than running on some separate stack. They use retrieval from the Search index to stay accurate and current. And both AI Overviews and the conversational AI Mode may issue multiple related searches across subtopics to assemble a single answer, which is why your page can get cited for one narrow aspect of a topic it only partly covers.
That last point matters more than most people realize, and I will come back to it.
The Three Findings That Should Shape Your Strategy
Strip away the noise and three things matter.
When an AI Overview appears, clicks decline. Study designs differ and the size of the drop varies enormously depending on methodology, query type and position. What is not in dispute is the direction. If you see a single dramatic number quoted without a methodology attached, be sceptical of whoever is quoting it.
Being cited is worth roughly double not being cited. On the same results page, a page cited inside the AI Overview earns meaningfully more clicks than an uncited page. Both still trail what you would have earned on a search with no AI Overview at all.
That gives you a clean hierarchy. Best outcome: rank on a query that does not trigger an overview. Second best: get cited inside the overview. Worst: rank on a page with an overview and not get cited.
Queries without AI Overviews are appreciating. This is the finding almost nobody acts on. Click-through rates on searches where no AI Overview appears have been rising, and that gap is widening rather than closing.
The explanation is straightforward. Overviews absorb the quick-answer queries. The people still clicking through on those searches wanted more than a summary, which makes them a higher-intent population.
Which means queries that do not trigger overviews are becoming a scarcer, more valuable asset. Most keyword portfolios have never been segmented this way. They should be.
What Google Actually Says
Worth quoting the substance accurately, because it gets misrepresented in both directions.
Google's position is that SEO best practices remain foundational for its generative AI features, because those features are rooted in core Search ranking and quality systems. There are no additional requirements to appear. No new files or special markup are needed. It has also said these features can surface a wider and more diverse set of links than classic web search, which creates opportunities for more types of sites to appear.
It has separately cautioned against third-party tools promising ranking success or claiming access to internal Google metrics, noting that no third-party tool has that access.
You can read both the AI features and your website documentation and the guide to optimizing for generative AI features directly.
Here is how I read it. Google is telling the truth about eligibility and a partial truth about outcomes.
"No special optimization is required to be eligible" is accurate. But eligibility is not selection, and selection is not click retention. A page can be perfectly eligible, get cited, and still see fewer visitors than it did two years ago.
Trust the guidance on mechanics. Ignore anyone selling a secret method. Plan for the traffic reality independently.
What I Would Actually Do
Segment the portfolio first
Before anything else, because you cannot make good decisions across an undifferentiated keyword list.
Build three lists. Queries that trigger an overview, where the goal shifts to citation and brand impression. Queries that do not, which you protect and expand aggressively. And intermittent ones, which is where changes show up first.
Then re-cut your reporting so traffic changes get attributed to the right cause. Most of the panic I encounter comes from teams comparing one aggregate number year over year without knowing which segment actually moved.
Restructure content for extraction
This is the highest-return content work available and none of it is exotic.
Answer in the first paragraph under the heading. These systems pull heavily from opening content. If your answer arrives in paragraph six after a scene-setting introduction, you have made yourself hard to cite.
Phrase your headings as the questions people actually ask. This matters enormously given that the system is assembling answers from many narrow sub-questions. Well-organized subheadings increase your chances of being cited for a specific subtopic even when you do not own the whole topic.
Increase factual density. Numbers, dates, units, specifics. A synthesis engine needs something concrete to attribute.
Publish something only you have. Original research, proprietary benchmarks, aggregated data from your own operations. Commodity content that restates what ten other pages already say has almost no citation value, and Google's guidance is explicit about the importance of unique, non-commodity content.
Date everything, and refresh honestly.
Get the technical floor right
Server-rendered content. Fast response times. Accurate structured data. Real author attribution with verifiable credentials. Clean internal linking. And bot rules that do not accidentally block the systems you want reading you.
Unglamorous, and non-negotiable.
Build corroboration outside your own domain
Generative systems behave like trust networks. They favour sources that other trusted sources already reference. Your own site is a claim. Third-party corroboration is evidence.
The large majority of brand mentions in AI answers originate on pages you do not own. Earned media, industry publications, review platforms, professional communities. Most organizations have this budget in the wrong department.
Rebalance toward what still converts
If informational clicks are compressing, the honest response is to shift effort rather than shout louder in the same place.
More investment in transactional, local and branded queries. More investment in formats an overview does not replace: tools, calculators, video, live events, community. More investment in direct channels you own outright.
Let me be blunt about this, because it is the strategically important point. If your business model depends on renting attention from an intermediary that has decided to answer your customers' questions itself, the answer is not to optimize harder inside that arrangement. It is to build a relationship the intermediary does not sit inside.
Change what success looks like
Stop leading with sessions. Report citation frequency across your priority queries, impression-to-click ratio segmented by overview presence, revenue per session, self-reported attribution, and brand search volume.
Brand search volume is the cleanest proxy for whether AI exposure is compounding into something. Watch it.
The Reframe I Would Offer Leadership
Every organization I speak with wants to know how to get their clicks back. That is the wrong goal, and chasing it wastes a year.
For twenty-five years, search functioned as a traffic utility. Content in, visitors out, at a reasonably predictable exchange rate.
That utility is being repriced, permanently, and no amount of optimization restores the old rate.
What search is becoming is a credibility layer. Your content's job is shifting from earning a click to establishing that you are the authoritative source, whether or not the click follows. The value shows up later and elsewhere. In brand searches. In direct traffic. In the shortlist a buyer arrives with. In the recommendation an assistant makes when somebody asks for options.
That is harder to measure and better to own. Traffic could always be taken away by an algorithm update. Being the recognized authority in your category is considerably more durable.
Organizations that spend this year trying to restore old traffic patterns will spend the year losing. Organizations that spend it becoming genuinely worth citing will find the position compounds.
That is not a tactic. It is what a durable business looks like when the distribution layer changes underneath it, which it has done roughly every decade since the internet existed and will do again.
Search disruption and AI visibility are themes I work through with executive audiences. Explore the keynote topics here, or read the FAQ for specific questions about what this means for your leadership team.
You might also want to read What Leadership Capabilities Matter Most When Your Organization Is Disrupted by AI? and How Fortune 500 Leaders Are Really Responding to AI Disruption.
Frequently Asked Questions
What are Google AI Overviews?
AI Overviews are AI-generated summaries that appear at the top of Google search results, built from multiple sources with supporting links. They are grounded in Google's core Search ranking systems and may issue multiple related searches across subtopics to assemble a single response, which is why a page can be cited for one narrow aspect of a topic.
Do AI Overviews reduce click-through rate?
Yes, though the size of the effect varies substantially by study methodology, query type, device and position. What is consistent across every credible study is the direction. Treat any single dramatic figure quoted without a methodology attached with caution.
How do I get my website cited in AI Overviews?
Rank well in conventional organic results first, since the overwhelming majority of citations come from pages already ranking. Then structure content for extraction: answer the question in the opening paragraph under each heading, phrase headings as real questions, increase factual specificity, publish something only you have, and keep content dated and refreshed. Then build third-party corroboration.
Does being cited actually help traffic?
Yes, substantially, relative to not being cited on the same results page. Citation roughly doubles your click rate compared to appearing uncited. Both still trail what the same query would have delivered with no AI Overview present, so citation mitigates the loss rather than eliminating it.
Is SEO dead because of AI Overviews?
No. Google's own documentation states that SEO best practices remain foundational because its generative AI features are rooted in core Search ranking systems, and that no additional requirements or special optimizations are needed for eligibility. What changed is the payoff structure, not the discipline.
What is Google AI Mode and how is it different?
AI Mode is a conversational search experience built for questions that need exploration, reasoning or complex comparison, where you can ask nuanced multi-part questions that would previously have taken several separate searches. AI Overviews are summaries appearing above conventional results. They may use different models, so the links each surfaces will differ.
Should I block Google's AI from using my content?
For most businesses, no. Blocking removes you from the surfaces your customers are actually using without removing the underlying behavioural shift. There is also a practical trap where rules aimed at AI can inadvertently affect general search crawling. Publishers with direct subscription models face a genuinely different calculation than most commercial sites.
How should I measure success if clicks are falling?
Shift the headline metric from sessions to citation frequency across your priority queries, impression-to-click ratio segmented by AI Overview presence, revenue per session, self-reported attribution captured at conversion, and brand search volume. Fewer visitors converting at higher rates reads as failure on a session dashboard and as success on a revenue one.