The AI-visibility glossary

An AI system that gives people a direct answer instead of a page of links. ChatGPT, Gemini, Perplexity, Copilot, and Google's AI Overviews are all answer engines. When a shopper asks one what to buy, it names a few products. Being one of them is the goal.

The practice of making your store the source an answer engine trusts and names. SEO aims for a high rank on a results page. AEO aims to be the shop the AI recommends. Read the full guide →

The long-standing craft of ranking higher in a list of blue links on Google or Bing. It's still useful. It's built for a page of choices, while AI gives a shopper one answer first. AEO vs GEO vs SEO →

Another name for the same idea as AEO: getting picked up by generative AI when it writes an answer. Some people separate the two, but most use them interchangeably. If someone says GEO, they mean showing up inside AI-written responses.

When a shopper gets everything they need from the answer itself and never clicks through to a website. It's great for the person asking. For stores, it means being mentioned in the answer counts, along with site visits.

Shopping that happens inside a back-and-forth chat rather than on a grid of product tiles. The buyer describes what they want in their own words, and the assistant narrows it down. It rewards stores whose facts clearly answer a plain-language question.

Facts about your products written in a format machines can read cleanly: price, availability, brand, materials, dimensions. To a person it looks like an ordinary product page. Underneath, structured data spells out the same facts so an engine doesn't have to guess.

The shared vocabulary most structured data is written in, agreed on by the big search and AI companies. Using its Product type to label price, brand, and availability is one of the most direct ways to be readable to an answer engine.

Whether an AI's bots can reach and read your pages. Fast, open pages with the important text in plain HTML are the easiest to read. What the crawler can read, the answer can quote.

A specific thing an AI recognizes as real and distinct: your brand, a product line, a material. When engines know your store as an entity, they can connect facts to you with confidence.

The web of entities and the relationships between them that an AI leans on to reason. Your brand sits inside it linked to what you sell, where you ship, and what you are known for. Strong, consistent links make it easier for AI to name you.

The same fact confirmed in more than one place. An engine trusts that your apron is organic linen far more when your page, a review, and a directory all agree. When several sources line up, the AI is more likely to repeat the fact.

How current your published facts are. Prices change, items sell out, lines get retired. Answer engines favor sources that stay up to date, so shoppers get what was recommended.

When an AI states something that is not true, like a price or a feature it invented. Clear, corroborated facts help most. When the real answer is easy to find, the model is more likely to get you right.

The shopping experience inside ChatGPT, where the assistant suggests specific products in the flow of a conversation. For many sellers it's the first place they see AI send real buyers their way. Shopify & ChatGPT →

Google's AI-written summary that sits above the classic list of links, answering the question directly. A related mode, AI Mode, turns the whole search into a conversation. Both choose which sources to pull from. Being one of them is the goal.

An answer engine built around citations: it responds in prose and shows the sources it drew from. Because it names where each fact came from, it's one of the clearest places to see your store used as a source.

Google's Gemini and Microsoft's Copilot are the assistants woven through the products people already use, from search to email to the desktop. Each can recommend products, so each is another place your store can be named.

The common way engines answer with current facts: before writing, they retrieve real sources, then generate a response grounded in what they found. Clean, readable pages that other sources back up are what gets retrieved.

The moment an engine names or links your store as a source behind its answer. It's a clear sign that an engine trusted you enough to quote you. Citations are what AEO is measured by.

Tying an AI's answer to real, retrieved sources instead of only its training. A well-grounded answer points back at pages it can defend. The better your facts are published, the more often answers can be grounded in your pages.

Shopping done by an AI agent acting for the buyer: it compares options, picks, and can even check out on their behalf. The person sets the goal and the agent does the legwork. Stores that software can read clearly are ready for it. Read the guide →

The agent itself: the piece of software that reads product facts, weighs them against what the buyer asked for, and makes a choice. It reads your facts, so clear, complete facts help it choose you.

An open standard that lets AI assistants and agents connect to outside tools and data in a consistent way. Think of it as a shared plug. It helps an agent reach a catalog, ask about stock and act, with no custom integration for each store.

A structured file of your catalog, kept current, that machines can read in one pass: titles, prices, availability, identifiers. Feeds have powered shopping channels for years. In agentic commerce they're one way an agent learns what you sell.

The global identifier on a product, the number behind its barcode. It lets an engine tell whether your item and a competitor's are the same thing. Clean identifiers help the AI match your exact product to the exact question.

How often you get named across a set of buyer questions, compared with everyone else answering the same ones. It's the AI-era version of share of voice, measuring how much of the answer is you.

The share of tracked questions where an engine cites your store. Watched over time, it shows how your visibility is changing.

The list of real questions you track: the ones your buyers would type. A good prompt set reflects how people ask, from broad ("best gift for a new cook") to specific ("machine-washable linen apron"). Everything else is measured against it.

Whether you show up at all for a given question, before you worry about how prominently. Presence is the first step. Once you appear, you can work on being preferred.

A snapshot of where you stand today, compared with your own past and with competitors. It lets you measure progress. With one, "we're doing better with AI" becomes a number you can point to.

A planned single read on how ready your store is to be found and named by AI, built from the things engines check: readable facts, corroboration, freshness and more. Each point ties to something you can fix. See the Score →

The way Herald works, as one continuous cycle: Pull your catalog, Enrich the facts, Publish AI-readable pages, Track real citations, then start again as things change. How Herald will work →

Filling the gaps in your catalog so an engine has enough to quote you with confidence: the missing material, the exact dimension, the care instruction. It gives an AI what it needs to build an answer from your page.

The AI-readable pages Herald publishes for your products, laying your enriched facts out in the clean, structured form engines read best. The pages are ready to use.

Herald's way of shaping how your store meets an AI agent. It's the same care a good storefront gives a human visitor, aimed at the software doing the shopping. Read about AUX →