Glossary
AI search terms, defined for stores
Updated
19 terms a merchant meets when AI engines start answering shopping questions. Each entry is one sentence of definition, then what it changes for an online store. The same sentences are used across this site and in PositionBird's own reports, so a word means one thing everywhere.
SEO (search engine optimization)
Search engine optimization (SEO) is the practice of earning a ranked link on a search results page for the queries a store's pages could answer.
SEO is the first of three visibility layers, and it still feeds the other two: the pages Google and Bing index are the pages answer boxes lift and AI engines retrieve. In one Seer Interactive analysis, 87% of ChatGPT search citations were pages ranking in Bing's top ten for the query, so a store that has never earned a ranking rarely earns a citation either.
Source: Seer Interactive, 87% of SearchGPT citations match Bing's top results (2025-02-06). · GEO vs SEO vs AEO
AEO (answer engine optimization)
Answer engine optimization (AEO) is the practice of structuring content so a search engine or assistant lifts it as the single direct answer to a question, in a featured snippet, a People Also Ask box, or a voice result.
For a store, AEO is the difference between ranking for "are weighted blankets worth it" and being the paragraph Google shows for it. The work is on-page: a question as the heading, a direct answer immediately below it, and facts (price, materials, shipping) stated plainly enough to be lifted. Google has said it does not need special markup for its AI features, so AEO is about the visible text, not the schema.
GEO (generative engine optimization)
Generative engine optimization (GEO) is the practice of earning mentions and citations inside AI-generated answers, rather than ranking links on a results page.
GEO is the third layer. When a shopper asks an AI engine what to buy, the answer names a handful of brands and links a handful of sources; a store that is in neither is invisible for that question. Most GEO work is off-page: being present and accurately described on the review sites, comparison articles and retailers those engines retrieve from.
AI search visibility
AI search visibility is how often, and how prominently, AI answer engines such as ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews name and cite a brand when a buyer asks a question its products could answer.
It is the store-level outcome GEO works toward and the thing rank trackers cannot see, because there is no results page to rank on. Because engines vary from run to run, visibility is best expressed as a frequency (named in 6 of 10 runs) rather than a position, and tracked over time.
AI Visibility Score
The AI Visibility Score is PositionBird's 0-100 measure of how often AI engines mention and cite a brand across its tracked buyer-intent prompts, averaged over repeated runs per scan.
The score weights citations at 50%, mentions at 35% and engine coverage at 15%, so being linked as a source counts for more than being named in passing, and being present on several engines counts for more than dominating one. Its purpose is comparison over time and against competitors, not a grade of the store.
Share of Voice is the percentage of AI recommendations for a set of prompts that name a given brand rather than its competitors.
A store can hold a healthy score and still lose Share of Voice if a competitor is named more often on the same prompts. Because AI answers typically name only a few brands, Share of Voice is closer to a zero-sum measure than a search ranking is: a competitor's gain is usually a store's loss on that prompt.
Mention
A mention is an AI answer naming a brand in its text, whether or not the answer links to any of the brand's pages.
Mentions are how a shopper hears about a store in an AI answer, so they are the recommendation signal. PositionBird matches the brand name and its aliases with word boundaries, so "Bones" matches "the Bones store" but not "backbones", and the merchant controls the alias list.
Citation
A citation is an AI answer listing one of a brand's own pages among the sources it drew on.
A citation is the stronger signal: the engine actively pointed the shopper at the store's page rather than just naming it. It is also the one that drives AI-referred traffic, because the shopper can click it. PositionBird matches citations by host, so shop.example.com counts for example.com and notexample.com does not.
Mention vs citation
A mention names the brand in the answer text; a citation links one of the brand's pages as a source; an answer can do either, both, or neither.
They come from different mechanisms. A mention depends on what the model knows and what the retrieved pages say about the brand; a citation depends on the store's own page being retrieved and used. Tracking them separately tells a merchant which lever to pull: third-party presence for mentions, retrievable pages for citations.
Ghost citation
A ghost citation is an AI answer that cites one of a brand's pages as a source without naming the brand anywhere in the answer text.
It is common: engines often use a page for a fact (a price, a spec, a definition) while recommending a different brand. The shopper sees a link they may never click and a recommendation for someone else. A store with many ghost citations has solved retrieval but not recognition, which usually points to how the brand is described on the cited pages and elsewhere.
Buyer-intent prompt
A buyer-intent prompt is a question a shopper asks when deciding what to buy, such as "best soy candles for gifts" or "is retinol worth it for beginners".
These are the prompts worth tracking, because they are the ones where a mention or citation can change a purchase. PositionBird suggests them from the store's catalog and lets merchants add their own; a good set mixes "best X" questions, comparisons ("A vs B"), and "is X worth it" questions.
Answer engine
An answer engine is a search product that replies with a synthesized answer and a short list of sources instead of a ranked page of links.
ChatGPT search, Perplexity, Gemini, Claude with web search, and Google's AI Overviews and AI Mode all fit the definition. The consequence for a store is that there is no position 4 to hold: a brand is in the answer or it is not, which is why visibility in these products is measured as a frequency rather than a rank.
AI Overviews
AI Overviews are the generated summaries Google shows above its classic results for some queries, built from pages already in Google's index.
A page has to be indexed and snippet-eligible to appear in one; noindex, nosnippet or max-snippet:0 remove it from AI Overviews as well as from classic results. There is no official API that returns AI Overviews, so PositionBird does not collect them and does not publish data about them.
AI Mode
AI Mode is Google's conversational search tab, which answers a question by fanning it out into many sub-queries and synthesizing the results.
Fan-out means a store can be retrieved for a sub-question the shopper never typed ("which soy candles are unscented?") and cited in the final answer. Pages that answer one narrow question well are the ones that get picked up; broad pages that answer nothing in particular tend not to be.
Grounding / retrieval
Grounding, or retrieval, is the step where an engine runs a live web search and feeds the results to the model before it writes the answer, instead of answering from training data alone.
Which index the engine grounds on decides which pages can be cited: ChatGPT search draws on Bing and OpenAI's own crawl, Gemini on Google's index, and Claude's web search on Brave Search. Answers without grounding come from training data and cannot be changed by editing a page; answers with grounding can, once the page is re-crawled.
llms.txt
An llms.txt file is a plain-text page map placed at a site's root that proposes to tell AI systems which pages matter.
No major engine has committed to reading it. In an Ahrefs study of 137,210 domains, 28% published an llms.txt and 97% of those files received no requests at all in May 2026. Shopify generates one automatically for every store, so there is nothing for a merchant to add; it is harmless, but it is not a visibility lever.
Source: Ahrefs, llms.txt study of 137,210 domains (2026-06-15). · What to set up instead
Structured data / JSON-LD
Structured data is machine-readable markup, usually JSON-LD, that labels a page's facts, such as a product's name, price and availability, for search engines.
It earns display eligibility in classic search (product rich results) and keeps prices and stock consistent for crawlers. It is not an AI-citation lever: Google's documentation for its AI features says there is no special structured data to add, and tests of the chat engines show they read the visible page, not hidden markup. Keep Product and Organization markup accurate; do not expect it to move citations.
Source: Google Search Central, AI features and your website (2025-12-10). · When AI misquotes price or stock
OAI-SearchBot vs GPTBot
OAI-SearchBot is the OpenAI crawler that builds ChatGPT's search index, and GPTBot is the crawler that collects training data; blocking OAI-SearchBot removes a site from ChatGPT search answers, while blocking GPTBot does not.
Many stores added a robots.txt rule for GPTBot in 2023 and assume it covers ChatGPT. It does not: OpenAI's documentation says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, and that GPTBot only governs training. A store that wants citations should allow OAI-SearchBot whatever it decides about GPTBot.
Source: OpenAI, Overview of OpenAI crawlers (2026-09-04). · Checking crawler access
First-touch attribution
First-touch attribution credits an order to the channel that first brought the buyer to the store, so an order placed days after a ChatGPT referral still counts as AI-referred.
AI referrals rarely convert on the first visit and often land on a product page with no campaign tag, so last-click reporting undercounts them. RevenueBird records that a visit arrived from an AI engine and credits an order to that engine if the first AI-referred visit was within the previous seven days, with no cookies or personal data.
How the terms fit together
SEO, AEO and GEO are three layers of the same job: SEO earns the ranked link, AEO earns the lifted answer, GEO earns the mention and the citation inside a generated answer. The engines that produce those answers are answer engines; the step where they read the web before writing is grounding; what they return is measured as mentions and citations; and PositionBird summarises the result as an AI Visibility Score and a Share of Voice against named competitors, using the method on the methodology page. The longer comparison of the three layers is on GEO vs SEO vs AEO.
See the terms measured on your store
Free plan available. Or run the free AI shopping scan without an account.
Get it on the App Store