Skip to content
Vouch
Join the waitlist
Menu

GEO and AI search glossary

Plain definitions of the terms used in generative engine optimisation and AI search visibility, from AI Overviews and citations to Threads and the Vouch Rate.

Last updated by Vouch

The six stages of a ThreadA buyer conversation moves through Discovery, Shortlist, Objection, Trust, Value and Decision. In this example the brand is named for the first three questions and dropped at Trust, when the buyer asks whether the brand can be trusted.One buyer conversation, six questionsStage 1DiscoveryBest options?Stage 2ShortlistCompare threeStage 3ObjectionDownsides?Stage 4TrustReliable?Stage 5ValueWorth it?Stage 6DecisionWhich to buy?Dropped hereExample: your brand is named for three questions,then a rival takes its place when the buyer asks about trust.
A Thread follows one buyer through six questions. Being named in the first answer is not being chosen: a brand can drop out later, at Objection or Trust.
How the Vouch Rate is madeEach conversation scores from 0 to 1: half for how many of the six stages the brand survives, half for being recommended at the end, and sentiment moves the score by up to a tenth. The Vouch Rate is the week’s average times 100.Stages survivedHalfDepth of mention: 0 to 6 stagesChosen at the endHalfThe final answer recommends youSentimentUp to a tenthWarmer or cooler, up or downVouch Rate0 to 100weekly average
The Vouch Rate: depth of mention (stages survived), being chosen at the end, and a small sentiment adjustment, averaged over the week’s conversations into one number from 0 to 100.

Thread

A Thread is one buyer’s conversation with an AI assistant, six questions long: Discovery, Shortlist, Objection, Trust, Value and Decision.

Vouch Rate

The Vouch Rate is one number from 0 to 100: the average score of a week’s completed buying conversations, where a brand scores for each stage it survives and for being recommended at the end.

Discovery (Thread stage)

The first stage of a Thread. The opening question: the tracked prompt itself.

Shortlist (Thread stage)

The second stage of a Thread. Narrow down and ask to compare the leading options from the previous answer for your use.

Objection (Thread stage)

The third stage of a Thread. Raise the typical doubts and downsides about the leading options.

Trust (Thread stage)

The fourth stage of a Thread. Ask about reliability, after-sales service, warranty and real owner reviews.

Value (Thread stage)

The fifth stage of a Thread. Ask whether the leading option is worth its price compared with the alternatives.

Decision (Thread stage)

The sixth stage of a Thread. Ask for one clear recommendation and where to buy it.

GEO (generative engine optimisation)

The work of shaping what AI assistants and AI search features say about a brand, so that they name it, cite it and recommend it in their answers.

Read what generative engine optimisation is and how it differs from SEO and AEO in GEO vs SEO vs AEO.

AEO (answer engine optimisation)

Optimising content to be used as the direct answer in search features, such as featured snippets and voice answers, rather than as one link in a list.

The three disciplines compared: GEO vs SEO vs AEO. For a checklist, read the practical guide to answer engine optimisation.

AI Overviews

AI-generated summaries that Google shows in its search results for some queries. Google describes them as helping people get to the gist of a complicated topic or question more quickly.

Google's own description is on its AI features and your website page. See also how to be visible in AI Overviews and AI Mode.

AI Mode

A conversational search experience in Google Search where answers are written by AI and buyers can ask follow-up questions. Google says it is particularly helpful for queries that need further exploration, reasoning or complex comparisons.

Google's own description is on its AI features and your website page.

Share of voice

How often a brand is named in AI answers compared with its rivals, across the same prompts, assistants and countries. In Vouch it is the brand's share of the brand mentions in the tracked answers.

Average position

Where a brand appears, on average, among the brands an AI answer names: 1 means it is usually named first. Answers that do not name the brand are left out of the average.

Citation

A source an AI answer links to or names as the place its information came from. Citations show which pages, yours or other people's, an assistant relied on.

Own-citation share is the share of citations in your tracked answers that point to your own site. See how Perplexity chooses sources.

Salvation rate

The share of a week's completed Threads whose final answer recommends the brand.

It sits next to the Vouch Rate; the methodology shows how both are worked out.

Killer question

The follow-up question in an AI buying conversation where the assistant stops naming a brand and starts recommending someone else. It usually targets the brand's weakest documented claim.

Read killer questions: where brands lose the AI sale, and try the objection prompts to find yours.

llms.txt

A proposed plain-text file at the root of a website, written in Markdown, that gives AI tools a short guide to the site and links to its most useful pages.

What it does and what the evidence says: llms.txt and whether it helps AI visibility. How to write one, with examples: how to write an llms.txt file. The free AI crawler checker shows whether your site has one.

Query fan-out

A technique where an AI search feature breaks one question into several related searches across subtopics and data sources, then uses the results to write its answer. Google says AI Overviews and AI Mode may use it.

Google describes it on its AI features and your website page.

AI crawler

A bot that fetches web pages for an AI company, to train models, to build a search index or to answer a user's question there and then. Sites allow or block each one by its user agent in robots.txt.

Check which AI crawlers can read your site with the free AI crawler checker. What each one does, and what blocking it changes: should you block GPTBot?

Prompt tracking

Asking AI assistants the same buyer questions on a schedule and recording which brands each answer names, in what order and with which sources, so changes show up over time.

Vouch tracks single answers every day and follows the conversation with Threads. Starting points: buyer conversation prompts.

Sentiment

How warmly or critically an AI answer talks about a brand. A change in sentiment over a conversation can matter as much as whether the brand is named.

In Vouch, the change in sentiment from the first to the last mention moves a conversation's score by up to a tenth. See the methodology.

Grounding

Basing an AI answer on sources fetched at the time, such as web pages or search results, rather than only on what the model learnt in training. Grounded answers usually cite their sources.

Templates

How these terms become one measurement is set out in the Vouch methodology.