The click stopped. The question didn’t.
A marketing director opens Google Search Console and sees the same story every week now. Impressions hold. Clicks slide. The query is still there. The searcher just never lands on the page anymore.
That’s not a ranking problem. Pew Research tracked 68,879 real Google queries and found people click through to a traditional result in 8% of searches when an AI summary appears, against 15% when it doesn’t. A 47% relative drop, measured, not modeled.
The traffic didn’t vanish. It moved into the answer itself. The question a marketer has to answer changed from “how do I rank” to “how do I get cited inside the thing that answers instead of me.”
What GEO actually means
Generative engine optimization, GEO, is the name that stuck for that second question. It sits next to a close cousin, answer engine optimization (AEO), that describes almost the same practice from the marketing side rather than the research side. Neither term has a single body that owns the definition, so the vocabulary is still settling in real time. What both terms point at is the same shift: get referenced inside a generated answer, not just indexed underneath one.
That’s a different game from SEO in three specific places.
The ranking factor changes. SEO counts links pointing at one URL. A generative answer draws from how consistently a source gets cited across the whole web, not from backlinks to a single page. Fractl’s Dan Tynski analyzed 8,090 keywords across 25 verticals, 22,410 unique domains, and found only 7.2% of domains appeared in both Google’s AI Overviews and standalone LLM answers for the same query. The two systems are reading the web differently and reaching different conclusions about who’s authoritative. Optimizing for one doesn’t automatically win the other.
The win condition changes. SEO wins when someone clicks. GEO wins when someone reads the citation and never needs to. That’s a real loss for a site built entirely on ad impressions or a lead form. It’s a real gain for a firm that measures brand recall, direct traffic, and whether a prospect already trusts the name before the first call.
The competitive window changes, and it’s changing at different speeds depending on which term. This is the part most explainers skip.
The window closes faster than the guides admit
Search interest in the phrase “generative engine optimization” is real and it’s already competitive. Search interest in “chatgpt seo” rose from roughly a 17-point relative-interest average across the first six months of a recent 18-month tracking window to roughly 53.5 across the most recent six, a 221% increase. Two adjacent terms, “llm seo” and “geo vs seo,” show effectively no measurable search interest at the start of that window and become established, real, recurring queries only in the recent months. People weren’t searching for this eighteen months ago. They are now.
Search difficulty, informational vs. commercial
Internal Taqtics keyword research, July 2026
That gap is the whole story in one chart. The broad, informational way of explaining GEO is already a real fight. The narrow, commercial way of describing “someone who does this for me” barely has a defender in the search results yet.
One live example makes the size of that gap concrete: a site with a domain rating in the 40s, nowhere near an enterprise-scale competitor, currently holds the top organic position for the commercial “generative engine optimization agency” search, a real buyer-intent term with meaningful monthly volume. It got there with one page, not a content farm. The commercial side of this category still has room for a site without a decade of accumulated authority behind it. That won’t stay true indefinitely.
What actually changes in the work
None of this replaces the fundamentals. A slow site, thin content, and no real authority signals will lose in an AI answer the same way it loses in a list of ten blue links. What changes is what gets rewarded on top of those fundamentals.
Citation consistency beats keyword density. An AI system reading the web is checking whether a claim about a brand shows up the same way across multiple independent sources, a directory listing, a news mention, an industry report, the brand’s own page. One page stuffed with a keyword doesn’t move that needle. Five independent sources agreeing on the same fact does.
Original data beats restated advice. A generated answer needs something to point at that isn’t a paraphrase of the same five articles everyone else already wrote. A firm publishing a real number nobody else has, measured, dated, sourced, gives the answer something to cite that a competitor’s rewritten listicle can’t offer.
Direct and branded traffic become the trust signal, not just the vanity metric. If people are searching the brand’s name directly, or landing on the site with no referring search at all, that’s a sign the citation already worked, whether or not that specific visit came through a search result.
The click isn’t coming back the way it used to arrive. The question is whether the brand is the one the answer names when it explains itself. That’s a different build than a decade of keyword-first SEO trained anyone for, and the buyer-intent side of it is still mostly empty.
References
- Pew Research Center. "Google Users Are Less Likely to Click on Links When an AI Summary Appears in the Results." July 2025.
- Search Engine Land. "How AI media partnerships influence your brand visibility in genAI: Research." Fractl/Tynski analysis, October 20, 2025.
- Search Engine Land. "Google's AI Overviews are hurting clicks, Pew study finds." July 2025.
- Internal Taqtics keyword and search-trend research, July 2026.