AEO and GEO Labels Collide as Google Says AI Search Is Still SEO

A 2018 idea becomes the industry's main concern

In early 2018, a Search Engine Watch article argued that voice assistants were turning search engines into "answer engines." It made that case because most voice queries return one spoken answer, and optimizing for that answer seemed to need a different approach from classic SEO.22 Eight years later, the label from that piece, answer engine optimization (AEO), has a rival in generative engine optimization (GEO). Both now drive a crowded market of guides, tools, and agency services. The core question has not changed since 2018: when a machine answers for you, how does a publisher or brand stay visible?

Where the term came from is still disputed. Jason Barnard, founder of the brand-intelligence firm Kalicube, says he coined "AEO" in 2017. The earliest independent record, Rebecca Sentance's Search Engine Watch article from February 2018, quotes Barnard but says only that the practice "has come to be known as AEO," without crediting anyone.22 That old argument over credit fits the present moment, because the field still has no agreed name. In the United States, "generative engine optimization" draws about 8,100 monthly searches, while "answer engine optimization" and "AI search engine optimization" each draw about 4,400. At least six other labels also get meaningful search volume.22

Google's position: this is still SEO

The most important recent development in this debate comes from Google, not from a vendor. Google's own guidance mentions AEO and GEO as terms people use for work aimed at visibility in AI search, then says that from Google Search's point of view, this work is still SEO.22 The company is also direct about tactics. It says structured data is not required for generative AI search, that no special schema.org markup exists for it, and that content does not need to be written in a special way for AI.2226

That puts Google in open disagreement with much of the AEO advice now on offer. Several industry guides still treat FAQ schema and 40-to-60-word "answer capsules" as central to AEO.2919 One audit of AI engines found Gemini itself telling users that AEO requires FAQPage or HowTo markup plus 40-to-60-word answer blocks, which conflicts with Google's published guidance.22 Practitioners who quote Google's line argue that a clear, quotable sentence on the page matters more than the markup around it.26

Academic research gives Google's view some indirect support. The peer-reviewed paper that defined GEO reported gains from tactics such as adding statistics and quotations.13 A later peer-reviewed benchmark, however, found most of those optimization methods largely ineffective.22 Vendor guides tend to cite the first finding and leave out the second.

In my view, "AEO" and "GEO" work better as marketing categories than as separate technical disciplines. Even vendors selling GEO tools concede that Google's AI Mode and Bing's AI draw only from their own search indexes, so without basic SEO there is nothing for AEO or GEO to work with.23 Agencies make the same point in their own terms: answer engines quote pages that a search step has already found, so ranking is still how a page gets considered.26

The traffic numbers agree on direction, not size

The new labels have spread because the traffic damage is real. How large it is depends heavily on who measured it.

The most-cited independent figure comes from Pew Research Center. Its browsing data shows users clicked a traditional result on 8% of visits with an AI summary, against 15% without one, and clicked a link inside the summary on just 1% of visits.26 Other measurements span a wide range. Amsive's study of 700,000 keywords found an average click-through drop of 15.49%.18 Ahrefs puts the decline for the top-ranked result at 34.5%.7 Seer Interactive reported a 61% drop for brands that were not cited in the overview.12 A newer controlled study attributed a 39.8% reduction in clicks to AI Overviews, nearly all of it on informational queries.8

Estimates of how often AI Overviews appear vary just as much. One trend guide cites studies ranging from 4.5% to 12.5% of queries.7 An arXiv study of 55,393 queries found AI Overviews on 13.7% of trending searches but 64.7% of question-style searches.17 Similarweb puts the share at more than 40% of US searches.8 The gap is mostly about method: the share depends on whether a study samples trending queries, question queries, or a vendor's tracked keyword set. Anyone who quotes one number as "the" AI Overview rate is choosing a sample.

At the level of the whole market, SparkToro's June 2026 research found that 68.01% of US Google searches in the first four months of the year ended without a click. The same research found the share of searchers who click at least once fell by 9.51 percentage points between 2024 and 2026.17 SparkToro also found that only 0.34% of searches in its panel went into AI Mode over those months, and concluded the tab "played only a limited role" in the drop in clicks.8 That matters because it puts most of the current damage on AI Overviews and the broader results page, not on Google's conversational mode, at least for now.

Branded queries are the new pressure point

The most recent signal may be the most important one for brands. DemandSphere reported that AI Overviews appeared on 80% to 90% of its tracked branded queries in late September, up from 26% at the start of the month. Google had announced no change as of early October.8 Search Engine Land covered the jump and noted that its effect on clicks is not yet known.8

That matters because branded search has been one of the few areas AI Overviews left largely alone. Amsive's earlier work found that only 4.79% of branded keywords triggered an overview, and that branded queries actually gained 18.68% in click-through rate when one appeared.8 If Google is now putting AI summaries on a brand's own name, a common piece of advice (build your brand because branded traffic is safe) needs another look.

Publishers are already cutting jobs

The effects on jobs are no longer hypothetical. Press Gazette reported a second round of 2026 redundancies at Future, which its chief executive blamed on "the decline in search." Ipsos iris data shows Tom's Guide's UK audience down 55% from July 2024.8 Separate analysis cited by one vendor says Google referral traffic to publishers fell 38% year over year.20

Losses also vary widely by sector. Estimates of click loss range from about 1% for recipe sites to more than 24% for parenting content and up to 30.15% for health portals.8 E-commerce has largely been spared: one analysis says only 4% of e-commerce searches trigger AI Overviews, down from 29% at launch.2

Where the optimism comes from

The main argument for AEO rests on two claims: cited sources do better, and AI-referred visitors convert better. Several analyses report that brands cited in AI Overviews get 35% more organic clicks and 91% more paid clicks than brands that are not cited.620 Conversion figures for AI referrals range from 4.4 times organic in Semrush's measurement to much higher multiples in other reports.1316

These figures need a skeptical reading. Many come from companies that sell AEO tools or services, and they rarely use the controlled designs behind Pew's work. Volume is also still small. Conductor's analysis of 1,215 enterprise domains found AI referral traffic made up only 1.08% of all website traffic, with ChatGPT supplying 87.4% of it.11 A high conversion rate on 1% of traffic does not replace double-digit losses on the other 99%.

The link between citations and rankings is also unclear. One analysis citing Ahrefs data says only 38% of pages cited in AI Overviews also rank in the top 10, down from 76% seven months earlier. It attributes this to Google's "query fan-out," in which one query is split into many sub-searches.3 Another report found that only about 14% of URLs cited in AI Mode also appear in AI Overview citations.6 Yet some industry data claims 99% of AI Mode citations come from the top 20 organic results.19 These numbers can only all be true if "ranking" refers to rankings for the many fan-out sub-queries, not the original query. That reading supports Google's claim that this is still SEO, done across a much wider set of queries.

Measurement is improving

One practical problem is starting to ease. Google says Search Console's generative AI performance reports have now rolled out to all sites worldwide.9 Earlier, SEOs complained that AI Overview and AI Mode activity was mixed into ordinary Search Console data with no way to filter it out.5 Google also now labels a searcher's chosen "preferred sources" inside AI responses. A Google product manager says people are twice as likely to click through to those sources.8

The takeaway

The debate Search Engine Watch identified in 2018 has come back with far more at stake. The evidence points to a few conclusions. Clicks are falling, and the size of the drop depends on the query type and on who is measuring. Google sees AEO and GEO as new names for SEO, and has said publicly that special markup and AI-specific writing are not required. Most of the AEO playbook (clear answers, original data, solid authority, technical access) is good SEO under a new label.

The most important change to watch this autumn is not a new acronym. It is the apparent spread of AI Overviews onto branded queries. If that holds, AEO stops being about visibility for informational content and starts affecting whether a company's own name still sends visitors to its website.