AEO vs SEO vs GEO for ecommerce: what changes, and where to put the work

Search engine optimisation (SEO) is the work of earning a place on a search results page: crawlable pages, useful content, and the signals that help a search engine understand and rank them. The shopper sees a list and chooses where to click.

Answer engine optimisation (AEO) came out of the featured-snippet and voice-assistant years, in trade-press use by early 2018, when a search could end in one spoken or highlighted answer instead of ten links. The aim was to be that answer. For a store today, AEO usually means being the product an assistant names when a shopper asks for something like it; the AEO glossary entry has the short definition.

Generative engine optimisation (GEO) is the newest of the three. The term comes from a 2023 research paper by Aggarwal and colleagues, presented at KDD 2024, which measured how often a source appears in answers written by generative AI search. Other labels, such as LLMO and “AI SEO”, describe much the same ground.

The GEO paper built a benchmark of about 10,000 queries and tested nine ways of rewriting a web page. Adding citations to sources, quotations and statistics raised a page's visibility in the generated answers by up to 40%, while keyword stuffing did little, and on Perplexity made pages less visible. Pages that started lower in the results gained most from adding citations.

Those findings are about web pages being cited in answers, and most of the benchmark's queries were informational rather than shopping questions. The paper did not test product feeds, product listings or shopping carousels. For a retailer, it supports writing buying guides with sources and specific facts; it says nothing about which product an assistant recommends. That depends on the product data the assistant can read.

Google says there are no additional requirements to appear in AI Overviews or AI Mode, and no special schema.org markup to add: a page has to be indexed and eligible to show with a snippet, the same as in search, and appearing is not guaranteed (its AI features guidance, last updated December 2025). It also asks for structured data that matches the visible page and Merchant Center information that is kept up to date.

OpenAI asks sites not to block its search crawler, OAI-SearchBot, if they want to appear in ChatGPT search.

So the foundations are common to all three names: product pages that crawlers may read, accurate titles, prices, availability and variants, Product structured data that agrees with the page, and a healthy Merchant Center account. A store that has those right has done most of the shared work, whatever it calls it.

The shared foundations are what two of Sighted's pilots consisted of. With ChristianShirts.Net the work was structured content and schema improvements, and AI citation coverage went from 0% to 27%. With Onboard Sportswear it was duplicate URLs consolidated, the site's first sitemap submitted, indexing gaps addressed and the products enriched; AI citation coverage went from 0% to 33% of tested responses, organic clicks rose 59% and impressions in Google's AI features rose 1,140%.

Those three Onboard figures are three different measures on one site: AI citations, organic clicks and Google AI-feature impressions.

1. One answer names a few products

A results page shows many listings and leaves the choice to the shopper. An assistant writes one answer and names a handful of products in it, chosen against the request. A product that is fourth on a results page still gets seen; a product left out of the answer does not.

2. Product feeds are a direct input

Assistants do not rely on crawling alone. OpenAI takes product feeds from approved partners, with nine fields required for discovery (its standard feed upload targets the US for now), and Shopify merchants' products reach ChatGPT through Shopify Catalog.

Microsoft's Copilot Checkout takes product data through a Microsoft Merchant Center feed, and Google's AI Mode draws on its Shopping Graph, which Google said in May 2025 held more than 50 billion listings. The feed is now part of this work, no longer only a paid-shopping task, and the ChatGPT troubleshooting guide lists the checks a product has to pass on each route.

3. Detailed questions need detailed attributes

A shopper can ask for a waterproof jacket under £150 that packs into its own pocket. That is matched against attributes and stated facts, not a keyword in a title. Google added six optional conversational attributes to Merchant Center, including questions and answers and related products, to help AI surfaces such as AI Mode; Google says they do not affect approval, and it documents no ranking effect either way.

4. Answers vary from one run to the next

A ranking is fairly stable from day to day. An assistant's answer can change between runs and between people: OpenAI says ChatGPT's shopping results use the request and its context, including memory and custom instructions where they are on. One check of one question proves little.

5. Measurement moves from positions to mentions

SEO reports positions and clicks. AI answers are measured by whether you are named or cited, how often, and against whom, and the platforms report less of it.

Search Console's generative AI performance reports, open to every site since 31 August 2026, count how often your pages appeared in AI Overviews and AI Mode; they report impressions, not clicks. GA4's AI Assistant channel groups visits from assistants such as ChatGPT, Gemini, Copilot and Grok, while traffic from Google's own AI Overviews and AI Mode stays in Organic Search.

Merchant Center's AI performance insights add share of voice in AI Mode and AI Overviews, but as of October 2026 Google's help page lists them for English queries in Australia, Canada, India, New Zealand and the US, not the UK.

For UK stores, the missing layer is a fixed set of realistic shopping questions, asked on each assistant on a schedule, with the products named and the sources cited written down each time. Keep that log separate from rankings, so a good month in search does not hide a bad month in AI answers.

Is AEO the same as GEO?

Mostly. AEO is the older term, from featured snippets and voice answers; GEO comes from research on generative AI search. For a store the work under both names is the same: accurate, complete product data in every place an assistant reads, and content worth citing.

Do you need a separate AEO or GEO agency?

Not by default. Google says its AI features need no special requirements beyond good SEO practice. What is genuinely new is product feeds to the assistants, richer product attributes and a way to measure mentions. Check whether your current team or tools cover those before buying a separate service.

Does schema markup help with AEO?

Product structured data helps systems read price, availability and identifiers, and it should match the visible page. Google says no special schema is needed for its AI features, so treat markup as a foundation rather than a lever.

Which matters most for an online store?

Product data. Search, answer engines and generative assistants all decide on the facts they can read about a product, and the store is the only source of those facts.

1. Keep the SEO foundations: crawlable, indexable product pages, structured data that matches the page, and no AI crawler blocked by accident in robots.txt.
2. Fill the product attributes that real shopper questions turn on (size, material, compatibility, use, care), from your own verified data, and send them to the feeds and Merchant Center as well as the page.
3. Write buying guides with sources and specific figures, the part of GEO the research supports, and link them to the products they discuss.
4. Measure each layer on its own terms: rankings and clicks for search, impressions in Search Console for Google's AI features, and a prompt log for the assistants.

The product data optimisation guide covers the second step in detail, and GEO vs SEO for ecommerce covers which gap to fix first when a product is missing from AI answers.

Sighted works on the product data that all three depend on. It scores every product on the fields assistants read, writes the missing ones from your own descriptions, options and photos, publishes approved changes to your store, Merchant Center and the feeds, by hand or automatically as you choose, and checks on a schedule (weekly on Pro, monthly on Growth) whether the assistants name your products.

The free AI visibility audit is the scoring step on up to 50 of your pages; the enrichment runs at catalogue scale and is the technology behind SO-ME, the consumer shopping platform. Buying guides and links stay with your content team.