Blog
SEO vs GEO Is the Wrong Question. Search Visibility Is One Job Now

In October 2025, Krishna Madhavan of Bing published a post about Bing's approach to providing answers in AI search. His opening words were "Whether you call it GEO, AIO, or SEO, one thing hasn't changed: visibility is everything", emphasizing that crawlability, metadata, internal linking and backlinks remain important to Copilot (which is powered by Bing's search index).
This was followed in May 2026 by Google publishing its own advice for websites on how they can feature in Google's generative AI powered search experiences (namely Google AI Overviews (the AI summaries in Google's results) and AI Mode (its conversational search)). Its advice?
"From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO."
In other words, for Google's generative AI search experiences, GEO = SEO.
What is GEO?
GEO, Generative Engine Optimization, refers to the actions you can take to increase the likelihood of your website being cited or referenced in generative AI powered experiences, such as ChatGPT search, Perplexity, Gemini, Google AI Overviews / AI Mode, Copilot, etc.
Does GEO replace SEO?
For Google AI Overviews / AI Mode, no. Google AI Overviews and AI Mode are powered by Google's existing search index and ranking systems, so if your website isn't optimized to be found, crawled and indexed for relevant queries in Google, you're not going to be considered as a source for its generative AI features. Google makes this clear on its Search Central page about its AI features, your page has to be "indexed and eligible to be shown in Google Search with a snippet" to have a chance of being a supporting link in AI Overviews or AI Mode. And in its May 2026 advice on optimizing for Google's AI features it states that "our generative AI features on Google Search are rooted in our core Search ranking and quality systems."
So, for Google and Bing, SEO fundamentals still matter, but is that all there is to optimizing to feature in Google AI Overviews and AI Mode (or Copilot)? Not quite.
What does ChatGPT search index?
Unlike Google and Bing, OpenAI has been less forthcoming with details about how ChatGPT search works, its launch blog post (October 2024) states that it "leverages third-party search providers".
So, although ChatGPT is outside the scope of Google's SEO statement above, it (along with other AI experiences, like Perplexity) are a legitimate rationale for why GEO might be considered a separate (albeit overlapping) skill from SEO.
What is the overlap between SEO and GEO?
Research by Ahrefs (published August 2025, n=15,000 long tail prompts*) suggests that, on average, only 12% of the URLs cited by AI assistants also ranked in Google's top 10 SERPs for those prompts. The overlap was lower for ChatGPT (~8%), and higher for Perplexity (28.6%).
*Vendor reported research.
What was the original GEO paper about?
The term "GEO" originates from the paper "GEO: Generative Engine Optimization" (Aggarwal et al., presented at KDD 2024, arXiv preprint published November 2023), which introduced GEO-bench, a benchmark of 10,000 queries to measure visibility in generative responses across nine different content modification strategies. Their generative search engine considered "only the top 5 sources" (retrieved from Google) as potential sources for each query, and prompted gpt3.5-turbo to generate responses from those. Their best GEO strategies improved visibility "by up to 40%", with smaller improvements (up to 37%) measured when using Perplexity as the generative search engine. (Importantly, they found that "[s]imple methods like Keyword Stuffing traditionally used in SEO don't perform well.")
Visibility is not traffic
It's worth bearing in mind that the paper's definition of "visibility" is not traffic, but rather the share of the response attributed to each source, as measured by position weighted word count and a model judged impression score. Their experimental setup also means that improvements were measured for websites already in a fixed top 5, so recent research (a non-peer reviewed preprint surveying 45 GEO studies, July 2026) rightly cautions that Aggarwal et al's. results "are valid within its experimental setting but conditional on a source already being present in a fixed context". (Despite this, the paper's abstract asserts that generative engines are "rapidly replacing traditional search engines like Google and Bing", interesting given its own system only fed gpt3.5-turbo sources from Google.)
How well do AI Overviews overlap with Google's SERPs?
An updated Ahrefs dataset from March 2026 (n=863,000 keywords) found that 37.9% of AI Overviews citations also ranked in Google's top 10 for the same keyword. A further 31.2% ranked 11-100, and 31.0% ranked beyond 100.
Google has shared that for AI Overviews and AI Mode it may use a process it calls query fan out, where it runs multiple related searches across subtopics and data sources to answer your query. It's likely this is why Ahrefs found a relatively high level of overlap between AI Overviews citations and rankings beyond the top 10, around 69% of AI Overviews citations still ranked in Google's top 100.
Your site might not currently rank highly for your target queries, but might well be highly relevant to a related question that Google is fanning out to. Investigating related questions/subtopics is nothing new for SEOs (see our guide to long tail keywords for more), but it does mean you can't just rely on traditional rank tracking software to identify opportunities.
What about AI Assistants?
Ahrefs linked the lower AI Overviews overlap to fan out. It's less clear why overlap between AI assistants and Google's top 10 was as low as it was. We'd be interested in seeing more research on why the overlap is as low as it is, especially given the implications for GEO strategies, for now it seems prudent for brands to investigate both how they perform for customer questions in Google and AI assistants.
Which queries should a brand work on for AI answers?
Start with queries the brand already pays to rank for plus other meaningful queries from Search Console (the queries a page already gets impressions for organically). These are approximations of sub questions an AI engine is likely to search. That's the queryDNA idea, and it's why we say we're simply better at search. Start with the queries.
Then expand out from queries that already convert (use the search terms report, it includes CPC and conversion data) to identify related questions those queries may lead to.
Before optimizing to appear in AI Overviews check if you already rank highly in blue link results for these queries.
Don't forget Google's guidance that creating many pages containing such content "primarily to manipulate rankings or generative AI responses in Google Search violates Google's scaled content abuse spam policy", focus on improving answers on pages that already merit visibility.
SEO vs GEO, where does it actually differ?
Crawler access for AI retrieval bots
Blocking decisions are easy to get wrong, especially when made by someone trying to do the right thing. If your security team blocks all AI agents to avoid being used for training, you may well exclude your brand from ChatGPT search answers.
From OpenAI's documentation on their crawlers, OAI-SearchBot and GPTBot:
"Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links."
"Disallowing GPTBot indicates a site's content should not be used in training generative AI foundation models."
From Perplexity's crawler documentation:
PerplexityBot "is designed to surface and link websites in search results on Perplexity. It is not used to crawl content for AI foundation models." Its user triggered fetcher "generally ignores robots.txt rules."
Google recently (August 31, 2026) rolled out a generative AI control to all websites in Search Console (opted in by default) to let sites opt out of appearing in Google's AI powered results. Google's guide to appearing in AI powered results from Google includes opting in to these results as a prerequisite and notes that opting out means you "won't receive any traffic or impressions from these features."
Each of these crawler controls should have a written policy and an owner in marketing.
Many of the tactics recommended to optimize for generative engines are good SEO practice, crawlability, clear headings and structure, demonstrating E-E-A-T, using accurate schema etc. However, Google rejects some of the tactics recommended for their own AI powered results:
From their May 2026 guide: "There's no requirement to break your content into tiny pieces for AI to better understand it." and llms.txt files are currently ignored by Google Search.
From their AI features page: "no special schema.org structured data that you need to add"
When advice is aimed at Google, we prefer to stick with their guidance, "Prioritize effective SEO strategies over 'AEO/GEO hacks'."
Measuring AI search visibility
Brands should focus on query level performance first and use citation counts as a sanity check.
AI Overviews affect click through rate, when AI Overviews appear traditional results see a reduction in clicks. Here are a couple of recent studies:
In March 2025, the Pew Research Center tracked the Google searches of 900 US adults and discovered that on searches displaying an AI summary, participants clicked a traditional result on 8% of occasions (compared to 15% when there was no AI summary). They clicked a website within the AI summary 1% of the time.
Ahrefs analyzed desktop click through rate data in February 2026 (vendor reported, correlational) and reported that click through rate for keywords featuring an AI Overview was on average 58% lower for the first position result.
Ahrefs looked at correlations between brand signals and visibility in AI Overviews in May 2025 across 75,000 brands. In this vendor reported study, web mentions of a brand's name had a correlation with visibility in AI Overviews of 0.664, compared to 0.218 for backlinks. They caveat that correlation does not mean causation and Google notes in its May 2026 guide that "seeking inauthentic 'mentions' across the web isn't as helpful as it might seem". That said, PR (securing relevant coverage in publications your customers read) may deserve more attention than it currently gets.
From Microsoft's October 2025 post about optimizing for AI search: "ranking still happens, but it's less about ordering entire pages and more about which pieces of content earn a place in the final answer." Structure content so that each section directly answers a question in the first sentence to make extraction easier.
What's the bottom line?
Query intelligence now includes monitoring AI answers as well as organic search results, ownership of this should sit with the person who owns SEO. There needs to be someone responsible for understanding which queries the business wants to be cited for, and which bots can be blocked.
We recommend monitoring visibility for a predefined list of queries relevant to your customers and tracking this regularly across different engines. We'd recommend being skeptical about any 'AI visibility score'. Where possible we'd also recommend checking if citations are accurate, see our post on ChatGPT and Perplexity for more.
According to Sundar Pichai in Alphabet's July 2026 earnings remarks, Google has brought AI Overviews and AI Mode together into a single Search experience, and AI Mode alone is past 1 billion monthly active users. This makes it even more important for marketers to have a finger on the pulse of AI search.
We're big fans of Microsoft's October 2025 post, it includes lots of useful advice on optimizing for AI powered search results. It's advice like this that could help GEO efforts justify spend.
That's why we think of it as one job. Same owner, same query data.