--- title: "GEO for ecommerce, measured | Elivate SEO" description: "Shopify already sends store catalogs to ChatGPT and Google AI Mode by default. What we found on 60 D2C stores, and what a study says about product copy." url: "https://elivateseo.com/insights/geo-for-ecommerce" published: "2026-09-26T00:00:00.000Z" --- GEO # GEO for ecommerce, measured Shopify already sends store catalogs to ChatGPT and Google AI Mode by default. What we found on 60 D2C stores, and what a study says about product copy. Ridam Agrawal founder, elivateseo **Published:** 26 September 2026 **Read:** 13 min In short ## What is GEO for ecommerce? GEO for ecommerce is getting a store's products into the answers ChatGPT, Google AI Mode and Gemini give shoppers. For Shopify stores, much of it is already on: Shopify sends the catalog to those channels by default, as of September 2026. What is left for the store is its product data, its reviews and its listing copy. GEO for ecommerce, or generative engine optimization for online stores, is the work of getting a store’s products into the answers AI assistants give shoppers. Seven of the eight pages ranking on Google for “geo for ecommerce” in September 2026 are written by software vendors and agencies, and between them they define the term and list tactics and tools. Four mention Merchant Center or product feeds in passing. None mentions Shopify Catalog or OpenAI’s product feeds, or that Shopify already sends store catalogs to ChatGPT, Google AI Mode, Gemini, Microsoft Copilot and Meta by default. This page covers those routes, what is already switched on, what we measured on 60 direct-to-consumer stores on 26 September 2026, and what the one ecommerce study on that results page found about product copy. The study ranks eighth, and none of the seven pages above it cites it. ## Where a product in an AI answer comes from A product reaches an AI shopping answer by one of four routes, according to the platforms’ own documentation as of September 2026. - **Shopify Catalog.** Shopify syndicates a store’s products to AI channels “with their title, description, options, images, price, availability, and other key attributes, all structured in a way that AI agents can parse and understand”, and keeps the data current. - **Google Merchant Center.** Google’s guide to generative AI features says Merchant Center feeds and Business Profiles “can help your products and services to be visible in both AI responses and other Google Search results”. Shopify routes products to Google AI Mode and Gemini through its Google & YouTube channel or Merchant Center. - **OpenAI’s product feeds.** A merchant can send ChatGPT a structured feed directly, but OpenAI says onboarding “is currently available to approved partners”. - **The open web.** Shopify notes that “AI crawlers might also access your store directly through the open web”, and OpenAI lists “other third-party content” among what ChatGPT weighs. The protocols in the news sit a layer above this. OpenAI and Stripe’s Agentic Commerce Protocol, and the Universal Commerce Protocol that Google’s guide names, handle checkout and platform integration. Implementing the Agentic Commerce Protocol does not list a store anywhere: its own FAQ answers that question “No”, and adds that a business wanting to participate in ChatGPT “will need to apply”. ## What is already switched on for a Shopify store Shopify’s agentic storefronts put a store’s products in ChatGPT, Google AI Mode and Gemini, Microsoft Copilot and Meta, and Shopify’s help centre says the feature “is active by default for eligible stores”. The controls are in the Shopify admin under **Sales channels > Agentic**. OpenAI’s help page says the same from the other side: for Shopify merchants, product data “is already integrated into ChatGPT through Shopify Catalog”, and “No additional work is required from individual merchants.” Shopify also says every store “automatically serves” three files written for AI agents, `/agents.md`, `/llms.txt` and `/llms-full.txt`, carrying the store’s name, sitemap, policies and discovery endpoints. None of the seven vendor and agency pages ranking for “geo for ecommerce” mentions them, so on 26 September 2026 we checked. We fetched the homepage, `/agents.md` and `/llms.txt` of the 60 direct-to-consumer brands from [our AI-crawler study](https://elivateseo.com/insights/who-blocks-ai-crawlers). 45 homepages answered. Shopify’s own response header identified 31 of them as Shopify stores, and Shopify’s CDN in the markup identified two more. | | Shopify stores | Other stores | | --- | --- | --- | | Stores that answered | 33 | 12 | | Serve a real /agents.md | 24 | 0 | | Serve a real /llms.txt | 28 | 3 | | Serve Shopify’s generated /agents.md | 22 | — | A file counted only if it opened with a Markdown H1, the test from [our llms.txt census](https://elivateseo.com/insights/llms-txt); an HTML page answering at the path did not count. Twenty-two of the 33 Shopify stores serve Shopify’s generated `/agents.md` unchanged, down to the file’s own line calling itself “the canonical agent-facing description of the store”. None of the 12 other stores serves one. For these brands, the platform decided whether an AI agent file exists, not the brand. Nine of the Shopify stores serve no real `/agents.md`, so “every store” is not what we observed, and why could not be seen from outside. The same stores’ product pages, read for Google’s merchant-listing markup, are in [D2C SEO, measured on 43 brands](https://elivateseo.com/insights/d2c-seo). ## robots.txt is not where stores lose AI visibility The standard GEO advice is to check that robots.txt does not block AI crawlers. For the direct-to-consumer brands we have measured, that is a problem almost none of them has. In our robots.txt study of 3 September 2026, 46 of 60 D2C brands answered. Forty-two never mention an AI crawler, the four that do mention one do so to allow it, and not one blocks a training crawler or a retrieval crawler. The 12 news publishers that answered the same study block retrieval crawlers, all 12. The access question for stores sits one layer lower, at the firewall. Fourteen of the 60 D2C brands would not serve robots.txt to our declared research bot on 3 September. On 26 September, 15 of the 60 would not serve their homepage to our audit tool’s declared bot: 11 refused with a 403 or 429 status and four did not answer at all. That is not evidence that those stores block ChatGPT or Google. Verified AI crawlers arrive from their operators’ published IP ranges, which a firewall can allow while refusing us, and what a real OAI-SearchBot receives from a store cannot be measured from outside. Shopify says bot management is handled for its stores, needs no action, and advises against putting a proxy in front of Shopify. A store that runs its own firewall in front of its site should check that the firewall and the robots.txt file allow the same crawlers. ## What E-GEO found about product descriptions E-GEO is the one ecommerce study on the “geo for ecommerce” results page: a paper by Puneet Bagga, Vivek Farias, Tamar Korkotashvili, Tianyi Peng and Yuhang Wu, first posted to arXiv in November 2025 and revised on 14 July 2026. The researchers took 13,747 shopping queries, each paired with 10 Amazon listings, rewrote one listing’s description, and measured how many places it moved in the rankings of five AI models: GPT-5, Claude Sonnet 4.5, Gemini 3 Flash Preview, DeepSeek V3.2 and Llama 4 Maverick. Results are the average movement out of 10 positions, on 2,000 test queries. The first test was the cheapest thing a store owner can do: ask a chatbot to “Optimize my product description so that an LLM is more likely to rank it higher than similar products.” Whether that helped depended on the chatbot. | Model doing the rewrite | Average places gained | | --- | --- | | Claude Sonnet 4.5 | +0.57 | | GPT-5 | +0.56 | | Llama 4 Maverick | +0.36 | | Gemini 3 Flash Preview | +0.26 | | DeepSeek V3.2 | +0.16 | | GPT-4.1 | −0.03 | | GPT-4o-mini | −0.36 | GPT-4o-mini made listings rank lower on all five models. Length made no difference in either direction: across 68,689 scored rewrites, the correlation between a description getting longer and ranking higher was about zero. GPT-5 shortened listings by about 14 words on average and ranked them higher; GPT-4.1 added about 170 words and ranked near the bottom. Bulleted lists carried a small positive effect. ## Which product copy styles lost rank in E-GEO The E-GEO researchers tested fifteen hand-written styles of the kind ecommerce guides recommend, with GPT-4.1 doing the rewriting and a plain rewrite scoring −0.03 as the baseline. Eleven of the fifteen did worse than the plain rewrite. | Style | Average places gained | | --- | --- | | Advertisement-like copy | −1.82 | | Foreign expressions | −1.66 | | Minimalist, one sentence | −1.49 | | Authoritative, “confident, assertive tone” | −0.64 | | Persuasive, “persuasive & compelling” | −0.62 | The authoritative and persuasive figures are our averages of the paper’s per-model table; the authoritative style lost 1.40 places on GPT-5 alone. Only four styles matched or beat the plain rewrite, all by small margins: imitating AI output (+0.14), adding an FAQ (+0.05), stating competitive advantages (+0.03) and formatting with headings and bullets (+0.02). When the researchers let an optimizer search for the best rewriting instructions instead, the optimized instructions converged on one playbook regardless of where they started, in the authors’ reading: an explicit ranking goal, the buyer’s intent, keywords and synonyms, an opening summary, section headings, scannable bullets, use cases, buyer questions, no keyword stuffing, and every original fact kept. The authors call it “a shared SEO-style ranking playbook”. GPT-5 and Claude also did not reward deliberately manipulative rewrites. The limits are the paper’s own. E-GEO is English only, its queries come from one Reddit community skewed to North American durable goods, its products are Amazon listings, and the ranking step is simulated by the models rather than measured on live ChatGPT or Google. ## What OpenAI and Google say they look at OpenAI’s help page for ChatGPT shopping says product results “are selected independently by ChatGPT and are not ads, nor influenced by any OpenAI partnerships”. It lists what ChatGPT considers: structured metadata from first- and third-party providers, other third-party content, the answer ChatGPT generated “before it considers any new search results”, and OpenAI’s policies. In OpenAI’s own example, ChatGPT “may consider available options, price, reviews, and ease of use”. OpenAI’s developer documentation is more direct: “recommended attributes—like rich media, reviews, and performance signals—improve ranking, relevance, and user trust.” Two consequences for a store follow from OpenAI’s page. ChatGPT “may generate simplified product titles and descriptions”, so the listing a store writes is not always the listing a shopper reads. And what ChatGPT already believes about a brand counts before it searches, which is where the rest of a brand’s presence on the web enters product selection. ChatGPT does carry product ads, and OpenAI’s feed specification keeps ads eligibility “independent of search eligibility”. Google gives stores one ecommerce-specific instruction, keeping Merchant Center and Business Profile current, and one general rule: “From Google Search’s perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO.” The rules are shared; the scoreboard is not. [Our GEO vs SEO measurement](https://elivateseo.com/insights/geo-vs-seo) found the sources cited by 65 AI Overviews overlapped page one of the same search by about a quarter. ## Measuring GEO for a store without buying a tool Ecommerce GEO has two free, first-party measurements, and neither appears on the pages ranking for “geo for ecommerce”. Zoovu’s page, ranking fourth, lists three AI-visibility tracking tools instead. Google Search Console’s generative AI performance report, which Google says it rolled out “to all websites worldwide” on 31 August 2026, shows how often a site’s pages appeared in AI Overviews and AI Mode, by page, country, date and device. The report counts impressions only, with no clicks, queries or positions, so it answers which pages were shown rather than for what. On Shopify, orders placed through AI channels “display in your Shopify admin with channel or referrer attribution”. That answers the question an impression count cannot: whether any of it sold. Between the two, a store can see which pages appear in Google’s AI features and which AI channels produce orders, without a third-party tracker. ## What to ignore Three things on the “geo for ecommerce” results page itself can be skipped. - **Chunking.** Google’s AI Overview for “geo for ecommerce” advises formatting content that language models “can easily chunk and quote”. Google’s own guide lists “‘chunking’ content” among the “AEO/GEO hacks” it says can be ignored. - **Shortlist sizes.** The same AI Overview says AI engines show “only 3 to 5 product recommendations”, citing a vendor’s YouTube video; one ranking page says ChatGPT gives “2–3 product names”. Neither figure is measured, and the two disagree. - **Recycled forecasts.** Three of the seven vendor and agency pages repeat Gartner’s 2024 forecast of a 25% fall in traditional search by 2026. One reports it as having largely happened, citing an SEO agency for that. ## What to actually do 1. **Open Sales channels > Agentic** in the Shopify admin and see which AI channels are on. Shopify switches them on by default for eligible stores. 2. **Read the store’s `/agents.md`.** Shopify wrote it; an `agents.md.liquid` theme template changes all three agent files. It will not move Google rankings, because Google Search ignores these files. 3. **Fix the product data the feeds carry**: titles, descriptions, options, images, prices and availability. Shopify’s Catalog Mapping handles data kept in metafields. Keep Merchant Center current for Google. 4. **Collect reviews.** OpenAI names reviews as a ranking input, and ChatGPT summarizes reviews from public websites. Gifted-product reviews carry a link rule, covered in [our off-page SEO guide](https://elivateseo.com/insights/off-page-seo-for-ecommerce). 5. **Write listings plainly**: what the product is, who it is for, specifications, use cases, buyer questions, and facts that are true. Skip ad copy and padding. A store that rewrites with AI should use a strong model; in E-GEO the weakest one made listings rank lower. 6. **Match the firewall to robots.txt** if the store runs its own firewall or proxy. The free [audit](https://elivateseo.com/audit) reports what robots.txt allows. 7. **Check the Search Console report monthly**, per page. ## What this does not show - **Not what AI crawlers receive from the stores that refused us.** A quarter of the D2C brands would not answer a declared bot; verified crawlers may still be allowed. - **Not that agent files do anything.** We measured whether the files exist. Google Search ignores them, and OpenAI’s documentation names feeds, not files. - **Not E-GEO’s numbers as effects on live ChatGPT or Google.** They are simulated rankings of Amazon listings, and the paper says so. - **Not why nine Shopify stores serve no agent file.** Headless storefronts, overrides or an incomplete rollout would each explain it. - **Not all stores.** The 60 brands are a named set of recognisable, mostly large D2C companies, not a random draw. This supports “of these stores” and not a share of all Shopify stores. A small store on a stock theme may look different. - **Not ChatGPT’s actual weighting.** The factors above are OpenAI’s description, quoted. ## Why we wrote it Seven of the eight pages ranking for “geo for ecommerce” in September 2026 are written by companies selling GEO services or ecommerce software, and so is every source Google’s AI Overview draws on for the same search. None of them cites the ecommerce study ranking beside them, and none mentions the platform documentation that answers the question a store owner actually has: where a product in an AI answer comes from. For most Shopify stores the answer is that the catalog is already there, sent by default. What is left is the part a store always controlled: its product data, its reviews and its listings, which the one study on the page says should read like good SEO copy with the facts kept true. ## Questions **Will GEO replace SEO?** GEO is not replacing SEO on the evidence so far. Google's own guide says optimizing for generative AI search "is optimizing for the search experience, and thus still SEO", and the 2026 E-GEO study found that product descriptions optimized for AI rankers converged on an SEO-style playbook: a summary, the buyer's intent, keywords, headings, bullets and facts kept true. **Do I need an llms.txt file for my Shopify store?** A Shopify store usually has one already. Shopify's help centre says every store automatically serves /llms.txt and /agents.md, and on 26 September 2026 we found a real /agents.md on 24 of 33 Shopify stores we checked, 22 of them Shopify's generated text. Google Search says it ignores llms.txt files. **How do I get my products into ChatGPT?** Products reach ChatGPT through structured product data from merchants and third parties, and through search. Shopify stores are connected through Shopify Catalog, and OpenAI says no additional work is required from individual Shopify merchants. Other stores can apply to OpenAI for direct product feeds, which OpenAI says are open to approved partners. **How do I measure GEO for an ecommerce store?** Google Search Console's generative AI performance report, rolled out to all websites on 31 August 2026, shows how often a site's pages appeared in AI Overviews and AI Mode, by page, country and device. The report counts impressions only, not clicks or queries. Shopify's admin separately attributes orders placed through AI channels. ## Sources - [Bagga, Farias, Korkotashvili, Peng and Wu: E-GEO, a testbed for generative engine optimization in e-commerce (arXiv 2511.20867)](https://arxiv.org/abs/2511.20867) - [OpenAI Help Center: Shopping with ChatGPT Search](https://help.openai.com/en/articles/11128490-shopping-with-chatgpt-search) - [OpenAI Developers: Agentic commerce, key concepts](https://developers.openai.com/commerce/guides/key-concepts) - [Agentic Commerce Protocol](https://www.agenticcommerce.dev/) - [Shopify Help Center: Shopify agentic storefronts](https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts) - [Shopify Help Center: Shopify Catalog and product discovery for agentic storefronts](https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/products) - [Google Search Central: Optimizing your website for generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) - [Search Console Help: Generative AI performance report](https://support.google.com/webmasters/answer/16984139) - [Elivate SEO: Almost nobody has decided about AI crawlers](https://elivateseo.com/insights/who-blocks-ai-crawlers) - [Elivate SEO: llms.txt, who publishes it and who reads it](https://elivateseo.com/insights/llms-txt) ## Ridam Agrawal Founder of Elivate SEO. Full-stack engineer who built the SEO and GEO platform this work runs on — keyword and rank tracking, site audits, and citation tracking across AI assistants. [More about the work →](https://elivateseo.com/about) ## Is your brand being cited? On a free 30-minute strategy call we look at where you stand in search and in AI answers, and what would change it. [Book a strategy call →](https://elivateseo.com/contact) keep reading ## Related reading SEO ### [B2B SaaS SEO, measured on 69 sites](https://elivateseo.com/insights/b2b-saas-seo) We checked 69 SaaS sites against the advice Google shows for "b2b saas seo". Most blog in a subfolder and build rival pages; 27 serve Markdown to AI agents. SEO ### [D2C SEO, measured on 43 brands](https://elivateseo.com/insights/d2c-seo) We checked 43 D2C brands against Google's own docs. The platform writes the product markup; stores skip shipping and returns, and FAQ rich results are gone. Service ### [What GEO actually involves](https://elivateseo.com/services/geo) Entity clarity, sources that agree with you, and measurement of where your brand is cited.