Google AI Overview and SEO: what actually changes and what to do

TL;DR: Google AI Overviews now show on roughly 13% of all searches, and they cut organic click-through rates by an estimated 34 to 64% on the queries they cover. Ranking beneath the answer is no longer enough. You have to be cited inside the Overview. That takes structured, directly answerable content, real expertise signals, and consistent citations across the wider AI search ecosystem.

What is Google AI Overview and why does it matter for SEO?

Google AI Overview (AIO) is the AI-generated answer block that sits at the top of many search results pages, above the blue links. Google rolled it out broadly in the US in May 2024. By early 2025 it was firing on an estimated 13% of all queries, according to tracking data from SE Ranking [1]. For health, finance, and how-to queries, the share runs considerably higher.

The SEO consequence is structural, not incremental. Ranking #1 used to mean your link was the first thing a user saw. Now the first thing many users see is a several-paragraph AI answer that already resolves the question. Google cites 2 to 8 sources inside that block. The rest of the organic results sit below it, often below the fold on mobile.

AIO turns a winning position into something closer to being the source of the answer instead of being the answer. Your page can rank #1 in the traditional results and still drive far less traffic than it did two years ago, because the Overview settles intent before anyone scrolls. This is a real change in how ai search works inside Google's own product.

Some queries never trigger an AIO. Navigational searches (brand names, login pages), most transactional searches, and some YMYL topics where Google stays conservative show them far less often. The population of affected queries is large. It is not everything.

How does AI Overview actually affect SEO metrics and traffic?

Traffic is the first thing marketers ask about. The honest answer: it depends heavily on query type, and the early data is not encouraging for traditional organic clicks.

SE Ranking analyzed 100,000 US keywords and found AI Overviews on roughly 13.14% of queries [1]. BrightEdge reported that AIOs had expanded across a large share of informational queries by mid-2024, with click-through rates on organic results below an AIO dropping sharply against SERPs without one [2].

The most specific CTR figure comes from a Seer Interactive analysis: for queries where an AIO appeared, organic CTR fell by an estimated 34% versus the same query set before the AIO showed up [3]. Other analyses put the range at 18 to 64%, depending on position and query type. Nobody has perfectly clean data here. The before-and-after is confounded by other Google changes running at the same time.

The direction holds across every source I trust. AIO presence correlates with fewer clicks to organic results. The pages cited inside the AIO do somewhat better, pulling clicks they might not have earned otherwise, but those citations don't fully cover the volume lost across the rest of the page.

Rankings still matter. Google pulls AIO citation sources heavily from pages that already rank well. A Semrush study found roughly 41% of URLs cited in AI Overviews also sit in the top 10 organic results for that query [4]. Traditional SEO is not dead. It's necessary and no longer sufficient.

Think of it as a two-layer game. Layer one is traditional organic ranking. Layer two is AIO citation, a separate optimization target that sits on top of ranking.

| Metric | Without AIO on SERP | With AIO on SERP | |---|---|---| | Avg organic CTR (position 1) | ~6-8% | ~2-5% (estimated) | | AIO citation sources | n/a | 2-8 URLs cited | | % AIO citations from top 10 | n/a | ~41% [4] | | Query share triggering AIO (US) | n/a | ~13% [1] |

How does Google decide which sources to cite in AI Overviews?

Google has not published a formal citation algorithm for AI Overviews. What researchers have reverse-engineered from large-scale analysis gives us a working model, and four factors keep showing up.

Authority and trust signals from traditional SEO carry over. High-authority domains get cited far more often. A Semrush analysis of more than one million AIO citations found a domain's overall authority score was a strong predictor of citation frequency [4]. Google's own E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) maps cleanly onto what the AIO selection seems to reward.

Content structure matters a lot. AI Overviews are generated answers, so Google's model is extracting information from your page to build a response. Clearly structured, directly answerable content is easier to pull from. FAQ sections, definition paragraphs, numbered steps, and tables all give the model clean extraction targets.

Topical completeness is real. Pages that answer the primary question plus the natural follow-ups in one place get cited more than pages that answer one thing narrowly. This is topical authority: covering a subject deeply enough that Google's model can lift several relevant facts from a single source.

Freshness weighs more than it used to. AIOs for volatile topics (news, product comparisons, regulations) tend to cite recently updated pages. A visible last-updated date, plus an actual refresh schedule, is no longer optional in those categories.

Schema markup helps. Pages with FAQ, HowTo, and Article schema give structured signals that appear in AIO citations at higher rates across several studies. Schema alone won't carry a weak page.

AI Overview trigger rate by query type (US, 2024)

| | | |---|---| | Informational queries | 42% | | All queries (avg) | 13% | | Commercial queries | 11% | | Transactional queries | 4% | | Navigational queries | 2% |

Source: SE Ranking, AI Overview study, 2024

How will AI change SEO strategy from here?

The core shift is from ranking to citation. It sounds subtle. It changes almost everything about how you brief content, measure success, and spend money.

Content strategy: the old model was keyword targeting plus link building. The new model adds a third pillar, being the source the AI model trusts enough to cite. That means writing content that answers questions directly and fully, not content built around keyword density. It means visible author credentials on the page. It means your page needs to be the most authoritative answer, more than an answer.

Measurement: if you judge SEO only by keyword rankings and organic sessions, you miss half the board. You need to track AIO citation frequency for your target queries. It's a first-class metric now. Several tools, including ai seo tools, offer AIO citation monitoring.

Link building: inbound links from authoritative domains still work as a proxy for trust, and trust drives AIO citation. Link building is not dead. But the marginal value of exact-match anchor text links from low-authority sites keeps dropping. Quality over volume was always good advice. It's basically mandatory now.

Brand signals: AI models, including the Gemini model powering AIOs, train on web-wide data and build implicit brand associations. A brand mentioned, cited, and linked across many authoritative sources becomes part of the model's trust graph in a way a brand that only appears in its own content never does. Earning brand mentions across press, Wikipedia, industry publications, and forums is an ai seo strategy now, not only a PR one.

The move from SEO to what practitioners call generative engine optimization is underway. It's not a replacement. It's a superset.

How will Google AI Mode change the Google SEO ecosystem?

AI Mode is Google's more aggressive follow-on to AI Overview. It entered limited testing in early 2025 and replaces much of the traditional SERP with a conversational interface running on a more capable version of Gemini [5]. Instead of a one-shot AIO above the links, users get a multi-turn conversation where the engine synthesizes from dozens of sources at once.

For the ecosystem, AI Mode is the scenario where traditional blue links become secondary. Google's own testing, described in its I/O 2024 announcements, showed users in these sessions submitted more follow-up queries and stayed on Google longer. Less traffic exits to outside sites per initial query. For publishers and brands, that compresses the window where an organic click even happens.

Google has said AI Mode draws citations from a wider source pool than standard AI Overviews, sometimes citing dozens of pages per session instead of 2 to 8. That could be an opening. Deep topical content gives you more chances to be cited across a multi-turn conversation than in a single AIO.

The google ai search shift touches structured data too. Google has signaled that rich results (star ratings, pricing, availability) still appear in AI Mode when the query is transactional. For ecommerce and local businesses, schema markup and Google Business Profile completeness stay relevant.

One honest caveat: AI Mode is still in limited rollout as of mid-2025. The ecosystem effects are directionally clear. The magnitude is not measurable yet. Anyone quoting precise AI Mode traffic numbers is extrapolating.

How does ChatGPT change SEO and how does it compare to Google AI Overviews?

ChatGPT and Google AI Overviews are separate systems with different citation mechanics, but they hand SEOs the same problem: your content may answer a user's question without that user ever clicking to your site.

ChatGPT with web browsing (GPT-4o and later) pulls live search results and synthesizes answers, citing sources inline. OpenAI hasn't published detailed data on its citation logic. Research from BrightEdge and others suggests ChatGPT web citations favor high-authority domains, well-structured content, and pages that rank well in Bing, since ChatGPT's web search draws partly on Bing's index [2].

So Bing SEO is no longer a rounding error. A page that ranks well on Bing is more likely to be retrieved by ChatGPT's browsing tool. Bing's ranking signals overlap heavily with Google's (authority, relevance, structured content), so optimizing one tends to help the other. Gaps still exist. Audit them.

The user base is large. OpenAI reported over 100 million weekly active users as of late 2023, and the number has grown since [6]. If even a fraction use browsing mode for research queries that touch your categories, the citation opportunity is real.

The difference from Google AIO is tone and shape. ChatGPT answers are conversational, not SERP-shaped. Its users are often doing deeper research, not quick lookups, which makes them more likely to click a cited source for detail. ChatGPT citation may drive higher-value traffic per citation than Google AIO, even if total volume is smaller.

To see your full AI search picture across ChatGPT, Perplexity, Gemini, and Google AIO, treat it as one ai search visibility problem instead of four. It's more accurate and cheaper to manage.

What content changes actually improve your chance of being cited in an AI Overview?

This is the tactical part. I'll be direct about what has evidence behind it and what is practitioner intuition.

Evidence-backed:

Write a direct answer in the first 40 to 60 words of each section. AI models are extractive when they build Overviews, and they lift from the top of a section, not the middle. Bury your key claim in paragraph four and it's less likely to get extracted. Not a new idea. It powered featured snippet optimization for years. It matters more now.

Use FAQ schema and write real FAQ sections. Google's AIO often quotes or paraphrases FAQ content because the structure maps directly to how Overviews get composed. A 2024 Ahrefs analysis found pages with FAQ schema were cited in AIOs at a meaningfully higher rate than comparable pages without it [7].

Put E-E-A-T signals on the page, more than imply them. Author bio with credentials, publication date, last-updated date, citations to primary sources inside the article. Google's quality rater guidelines define these signals explicitly [8], and they feed the model's trust scoring.

Be the primary source, or cite one. Pages that link to original research, government data, or clinical studies get cited more than pages that make claims with no sourcing. Google's model appears to run a rough form of citation quality assessment.

Practitioner intuition (less confirmed):

Longer content seems to win citation more often, but length isn't the cause. Longer pages tend to carry more topical coverage, more internal links, and more extractable facts. Write thorough content because it's thorough, not to hit a word count.

Freshness has an outsized effect on citation for time-sensitive topics. Refreshing a page's statistics, examples, and dates may matter more than bolting on new sections.

How should you measure AIO impact on your own site?

Google Search Console does not report AIO impressions or citation events as a separate data type as of mid-2025. That's a real gap. Here's what you can do.

Compare CTR trends for high-impression informational queries against a baseline from before May 2024. A CTR drop on queries you still rank for, with no ranking change, is a signal that AIO may be intercepting clicks. Filter to queries where you average position 1 to 5 to isolate the effect cleanly.

Use a third-party AIO monitoring tool. Several now track whether specific keywords trigger AIOs and whether your domain shows up as a citation source. This is the most direct measurement available. AI visibility tools are standard infrastructure for SEO teams at mid-market companies and up.

Track branded versus non-branded traffic separately. AI Overviews hit non-branded informational queries hardest. If your non-branded organic traffic has fallen since mid-2024 while branded traffic holds steady, AIO is a plausible cause worth investigating.

Spawned's own visibility tools (full disclosure: this is the vendor publishing this article) let you audit which queries trigger AIOs and where your domain sits in the citation pool against competitors. If you want to see it for your domain, a free AI visibility audit takes about five minutes.

Monitor Bing and Google separately. Bing's generative search has its own citation patterns, and its data feeds ChatGPT, so treating Search Console as the whole picture leaves you blind to a growing slice of the market.

Which industries and query types are most affected by AI Overviews?

The AIO trigger rate is nowhere near uniform. SE Ranking's 100,000-keyword study found informational queries trigger AIOs at roughly 3 to 4 times the rate of commercial queries, and transactional queries (buy, price, where to buy) trigger them least often [1].

By vertical, health and medical queries, how-to content, educational content, and financial explainers see some of the highest trigger rates. These are also the categories where Google cares most about source quality, so the trust bar sits high.

Local queries (near me, in [city]) behave differently. Google tends to serve its local pack and Google Business Profile results for these, with AIOs showing up less often. Local SEO is more insulated from AIO disruption than national informational content.

Ecommerce product pages are relatively protected on transactional queries. AIOs for product searches often include product carousels pulled from Google Shopping, a structured-data placement rather than a content one. Product schema and Google Merchant Center feeds matter here.

SaaS and B2B companies producing comparison content, best-of lists, and software reviews carry the highest exposure. These are exactly the informational query types where AIOs appear most, and they're the pages that historically drove high-intent traffic.

The ai-powered search features landscape moves fast enough that no industry analysis older than six months should be treated as current.

What is generative engine optimization and how does it differ from traditional SEO?

Traditional SEO optimizes for a ranking algorithm that uses links, authority, and relevance to order a list of results. Users browse and click. Generative engine optimization (GEO) optimizes for a generative model that synthesizes from many sources at once to produce a single answer, citing selected sources inline.

The inputs differ. In traditional SEO, a page wins by having the best authority-weighted relevance for a query. In GEO, a page wins by being the most trustworthy, extractable, and complete source for the information the model needs to build its answer. Heavy overlap. Not identical.

GEO-specific moves that don't map cleanly to traditional SEO: writing in a format that's easy to extract (clear claims, defined terms, structured data), building brand entity recognition across the web so the model's training data treats your brand as an authority, and tracking AI citation frequency as a direct metric instead of inferring it from CTR.

GEO also widens the optimization surface beyond Google. ChatGPT, Perplexity, Claude, and standalone Gemini each have their own retrieval and citation mechanics. A brand cited consistently across multiple engines is more resilient than one optimized only for Google's AIO.

A full treatment of GEO strategy lives at generative engine optimization. Short version: GEO is additive to SEO, not a replacement, but it needs a different measurement framework and some different content disciplines.

What should your SEO roadmap look like for the rest of 2025?

A few concrete priorities, ordered by expected impact.

Audit your highest-traffic informational pages for AIO citation. Find out whether your pages get cited when their target queries trigger an AIO. If they don't, identify the pages that do and reverse-engineer what they do differently. This gap analysis is the highest-leverage diagnostic you can run right now.

Refactor content to lead with direct answers. Prioritize pages that rank well but aren't cited. The fix is usually structural: move the answer to the top of each section and add an explicit FAQ block. Lower effort than net-new content, faster impact.

Build E-E-A-T infrastructure. If your pages lack visible author credentials, sourced statistics, and clear publication and update dates, fix that systematically. It affects both AIO citation odds and Google's traditional quality scoring.

Expand your brand footprint beyond your own domain. Guest contributions, press mentions, Wikipedia inclusion where warranted, and citations from industry publications all build the web-wide entity signal AI models use to judge brand authority. It takes months to accumulate. Start now.

Track AI citation as a KPI alongside traditional rankings. You cannot manage what you don't measure. Add AIO citation tracking to your monthly reporting. The ai search visibility metrics that matter in 2025 differ from the ones that mattered in 2022.

Don't panic over ranking drops that predate AIO. Some traffic declines since 2023 trace to the Helpful Content updates, some to AIO, some to broader SERP changes. Diagnosing which is which matters, because the fixes are different.

Sources

1. SE Ranking, AI Overview study (100,000 US keywords, 2024)
2. BrightEdge, AI Search Research Report 2024
3. Seer Interactive, AI Overview CTR Impact Analysis 2024
4. Semrush, AI Overview Citation Analysis 2024
5. Google, Google I/O 2024 keynote announcements and AI Mode documentation
6. OpenAI, usage statistics announcement (November 2023)
7. Ahrefs, AI Overview content and schema analysis 2024
8. Google, Search Quality Evaluator Guidelines (E-E-A-T)
9. Pew Research Center, Search Engine Use 2023
10. Google Search Central, Structured Data documentation