GEO for brands: how to get found when the shopper stops typing into Google

OpenAI rolled out shopping research inside ChatGPT in late 2025, describing it as a feature that researches deeply across the internet, reviews sources, and builds a personalized buyer’s guide instead of a list of links. The shopper describes what they want, ChatGPT asks clarifying questions, and it returns a short set of recommendations pulled from what it judges to be reliable retailers.

That single feature captures the shift. According to OpenAI’s own economic research, roughly 2% of ChatGPT queries involve shopping, which on a base that crossed 900 million weekly active users in February 2026 works out to tens of millions of shopping conversations a day. Google, Perplexity, and Amazon’s own Rufus assistant (now called Alexa Shopping in the US) are moving in the same direction, each building a discovery layer that sits between the shopper and the retailer’s website.

The commercial stakes are being taken seriously well beyond marketing circles. McKinsey has projected agentic commerce could generate $900 billion to $1 trillion in US retail revenue by 2030, and $3 to 5 trillion globally, while Google announced its own agent commerce protocol in January 2026 with Walmart, Target, and Shopify already signed on as partners.

Traditional SEO optimizes for position in a list, measured in clicks and traffic. GEO optimizes for inclusion in a synthesized answer, and that answer usually names three products, not thirty. The criteria an AI model uses to decide who makes that shortlist are different enough from classic ranking factors that they deserve their own checklist.

GEO checklist: what to fix to get recommended

Priority order: feed and crawlability first, then freshness, then off-site authority, then multi-engine tracking.

Structured data is the entry ticket, not a nice to have. Product feeds with complete Schema.org markup, accurate pricing, and real time availability are what let an AI system parse a catalogue with confidence. Pages carrying proper structured data are cited roughly three times more often in Google AI Overviews than pages without it, and OpenAI’s own Agentic Commerce Protocol expects feeds refreshed as often as every fifteen minutes.

Crawlability comes right after. If OAI-SearchBot, Google’s crawlers, or Perplexity’s bots cannot reach a page, nothing else on this list matters. Some retailers have made the opposite choice deliberately. Amazon has blocked ChatGPT’s shopping crawlers in its robots.txt, a defensive move that protects its own advertising business but also means Amazon listings simply cannot surface inside ChatGPT’s shopping results, leaving that visibility open to any brand selling directly or through a marketplace that allows the bots in.

Third party authority may be the piece most marketers underestimate. Generative engines lean heavily on independent sources to validate a claim before repeating it. One analysis by Yext found that 91% of AI citations trace back to third party content rather than brand owned pages, meaning comparison articles, review sites, and forum threads often carry more weight with an AI model than the product page itself.

Freshness and consistency close the loop. A price that differs between the feed and the landing page, or a stock status that has not been updated, is read by these systems as a trust signal failing in real time, and it can be enough to drop a product out of consideration even when the content itself is strong.

The honest answer is: mostly in the plumbing. A brand can have excellent product descriptions and still be invisible to an AI shopping assistant because its feed is missing half its attributes, its crawlers are blocked by an outdated robots.txt rule, or its pricing data lags behind the live site by a day. GEO starts as a data hygiene exercise before it becomes a content one, which is precisely why feed management and generative visibility are turning out to be the same conversation for e-commerce teams.

The category also matters. OpenAI has said its specialized shopping model performs particularly well in detail heavy categories such as electronics, beauty, home and garden, and sports and outdoor, exactly the kind of catalogues where product specifications, use cases, and comparisons genuinely help a buyer decide. Categories with dense, technical, comparison friendly content have the most to gain, and the most to lose if their data is not ready.

Start with the feed. Complete attributes, accurate pricing, live availability, and full Schema.org markup are the baseline every AI system checks first, and they’re also what marketplaces and comparison engines have always wanted, so the work holds up even if GEO turns out to be a passing label.

Google has made that literal. Alongside its move into conversational shopping through AI Mode and the Gemini app, it now reads a set of discoverability attributes straight from the standard product feed, fields like question_and_answer, related_product and popularity_rank that give Gemini enough context to reason about a product instead of matching keywords to it. They sit inside the same Google feed a merchant already runs, which is the clearest signal yet that AI visibility begins in the feed, not on the page. Lengow already supports these attributes, with Google’s native checkout following in early access.

Check what the bots can see. A quick audit of robots.txt against OAI-SearchBot, Google-Extended, and PerplexityBot tells a brand in minutes whether it has accidentally shut the door on an entire discovery channel.

Invest outside the owned website. Since the majority of citations come from third party sources, a strong presence in comparison content, review platforms, and trade press does more for AI visibility than another round of on-page tweaks. This is where digital PR, well sourced data studies, and genuine expert content earn their keep twice over, once for a human reader and once for the model reading on their behalf.

Track more than one engine. ChatGPT, Google AI Overviews, and Perplexity each pull from different sources and weight signals differently, and a brand optimized only for one is building on a single point of failure in a market that is still actively shifting.

GEO is not a rebrand of SEO with a new acronym attached, and it is not a fad either. It reflects something structural: an increasing share of the shopping journey is happening inside a conversation the brand cannot see, with a system that will only ever name a handful of options out loud. Winning a spot on that shortlist depends less on persuasion and more on giving AI systems clean, complete, trustworthy data to work with, at the exact moment they are looking for it.

For brands still treating this as a future problem, the window to act is open but narrowing. Early movers are already compounding an advantage that gets harder to close every quarter the gap widens.