---
title: "The AEO Vendor Glossary: 22 Terms That Measure Results or Change Them"
description: "A glossary of 22 AEO vendor terms, each tagged Measures or Changes, to help agencies tell a tracking dashboard from a platform that ships the work."
url: "https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/fc3582c6-c231-4152-9fdb-ab19e4c13eec.png"
published: "2026-10-12T05:20:11.005Z"
modified: "2026-10-12T05:20:11.068Z"
---

October 12, 2026 · 11 min read

# The AEO Vendor Glossary: 22 Terms That Measure Results or Change Them

[Alexander PerelmanHead Of Product @ groas](https://groas.com/author/alexander-perelman)[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![A wilting plant covered in sensors and gauges all reading red, while a dusty watering can sits unused: monitoring a problem isn't fixing it.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/fc3582c6-c231-4152-9fdb-ab19e4c13eec.png)

In this article

1. [Prompt tracking](#prompt-tracking)
2. [Tracked prompt caps](#tracked-prompt-caps)
3. [AI share of voice / visibility score](#ai-share-of-voice--visibility-score)
4. [Citation vs mention](#citation-vs-mention)
5. [Edge proxy](#edge-proxy)
6. [robots.txt and AI crawler access](#robotstxt-and-ai-crawler-access)
7. [llms.txt](#llmstxt)
8. [Schema markup and entities](#schema-markup-and-entities)
9. [Knowledge graph](#knowledge-graph)
10. [Answer-first content](#answer-first-content)
11. [Passage optimization](#passage-optimization)
12. [Direct CMS publishing](#direct-cms-publishing)
13. [Content gap](#content-gap)
14. [AEO vs GEO](#aeo-vs-geo)
15. [White-label](#white-label)
16. [Done-for-you vs self-serve](#done-for-you-vs-self-serve)
17. [Flat-rate multi-domain pricing](#flat-rate-multi-domain-pricing)
18. [Citation building](#citation-building)
19. [E-E-A-T and third-party corroboration](#e-e-a-t-and-third-party-corroboration)
20. [Entity consistency](#entity-consistency)
21. [AI crawler monitoring and referral traffic](#ai-crawler-monitoring-and-referral-traffic)
22. [Citation-rate guarantees](#citation-rate-guarantees)

An AEO dashboard can show you every prompt your client loses and still ship nothing that wins the next one. I read vendor pages with one question in mind: **does this term describe a measurement, or work that could change the result?**

Most of the vocabulary describes monitoring: prompt tracking, share of voice, visibility scores, citation rates. Useful, but none of those gets a client cited on its own. Fewer terms describe work: edge proxy, CMS publishing, schema and entity fixes, citation building. I tag each definition **Measures** or **Changes** so an agency shopping for white-label AEO can tell a dashboard from an execution system. When a rep blurs the line, ask who clicks publish. If the answer is your team, budget for your team to do the work.

## Prompt tracking

Prompt tracking runs a fixed set of buyer questions through AI search tools and records whether your client appears in their answers.

**Measures.** It records an appearance in an answer, not a URL ranking. [HubSpot frames a working cadence as 50–200 prompts run weekly](https://blog.hubspot.com/marketing/aeo-prompt-tracking) across engines including ChatGPT, Perplexity, and Gemini. Vendors misuse the chart when they sell it as progress. I used to tell clients tracking was step one. I was wrong about how long you can stay there. Tracking tells you which prompts you lose; it does not write the paragraph that wins the next run. **Buy the tracking only if you have a way to act on it.**

## Tracked prompt caps

A tracked prompt cap limits how many prompts a tool will run for you each month.

**Measures.** It also sets the price of that measurement. [Standalone HubSpot is $50/month for 25 prompts and 2,500 answers, with $20/month per extra 10 prompts, while Semrush charges $99/month per domain for 25 prompts and $60 per extra 50](https://seoprofy.com/blog/aeo-tools/). Split 25 prompts across two products and three funnel stages and you are watching a sample, not coverage. Price the cap against **prompts per client per week** before you price the retainer.

## AI share of voice / visibility score

AI share of voice compares your brand mentions with competitor mentions, while a visibility score counts how often you appear across tracked answers.

**Measures.** [Share of voice is your mentions divided by total brand mentions, yours plus competitors, times 100; Visibility Score is answers mentioning you divided by all answers tracked](https://blog.hubspot.com/marketing/aeo-prompt-tracking). What the deck calls _synergy_, I call two numbers doing different jobs. One tells you whether you are losing to someone specific. The other tells you whether you show up at all. Demand the raw counts: prompts tracked, answers with you, answers with them. **No denominator, no useful score.**

![Cartoon dashboard covered in dials beside a worker publishing pages](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/1be0168a-a042-46a9-ab20-52535810fd98.png)

## Citation vs mention

A mention names your brand in an AI answer; a citation gives the reader an attributed source to click.

**Measures.** [A citation may appear as a source link, footnote, or source card](https://netzender.com/aeo-mentions-vs-citations-key-differences-explained). Mentions can help you see whether the brand is recognised, but citations give you a link whose traffic you can track. [Rising mentions without rising citations can point to a missing citable page](https://netzender.com/aeo-mentions-vs-citations-key-differences-explained): perhaps the answer is thin, the name is inconsistent, or there is no clean passage to lift. Log both separately per query. Then fix the page instead of celebrating the mention chart.

## Edge proxy

An edge proxy sits between a client site and its visitors or crawlers, changing the HTML they receive without changing the CMS.

**Changes.** It can inject answer blocks, schema, internal links, and metadata. When a vendor says _no developer needed_, this may be the mechanism they mean. Otherwise, they may mean your team does the work by hand. Client development queues can stall a retainer over something as small as title tags; [an edge proxy gives agencies another way to ship those fixes](https://groas.com/post/how-an-edge-proxy-lets-an-agency-fix-15). Ask who approves the diff and how you roll it back. **A proxy is execution only when changes actually ship.**

![Cutaway diagram showing an edge proxy layer changing a page before crawlers read it](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/abb502d7-226c-4bb8-83a9-28b5cc92bc47.png)

## robots.txt and AI crawler access

robots.txt tells specified crawlers which parts of a site they may fetch.

**Changes.** Editing access rules changes what those crawlers can reach; a tool that merely flags a blocked crawler only measures the problem. Per-user-agent rules for GPTBot, ClaudeBot, OAI-SearchBot, PerplexityBot, and Google-Extended matter here, and [Google-Extended governs Gemini and AI use, not regular Googlebot search crawling](https://pagecrawl.io/blog/llms-txt-robots-txt-ai-crawler-monitoring). Before buying another visibility chart, [check raw HTML and crawler access with a curl and CDN audit](https://groas.com/post/can-chatgpt-actually-read-your-site-a-45). **If the vendor cannot show you the fetch, ask what its chart is based on.**

## llms.txt

llms.txt is a Markdown file at a site’s root that curates pages and summaries for models to use.

**Changes, barely.** It is neither robots.txt nor a sitemap: [robots.txt controls what crawlers may fetch, a sitemap lists what exists, and llms.txt curates LLM-friendly material](https://searchengineland.com/llms-txt-isnt-robots-txt-its-a-treasure-map-for-ai-456586). It grants nothing, blocks nothing, and no crawler has to read it. Vendors misuse it when they present the file as a substantial deliverable on its own. Publish it if you want; **do not mistake the file for the work that gets cited.**

## Schema markup and entities

Schema markup is structured page data that identifies things such as an organisation, an article’s author, or a product.

**Changes.** JSON-LD can make those identities explicit, but counting schema types added is not the same as fixing errors. The priorities I would ship first include [FAQPage, HowTo, Article with author and dates, Organization with sameAs, Product, and Speakable, implemented as JSON-LD and validated](https://www.rankshift.ai/blog/best-answer-engine-optimization-strategies/). One valid Organization block beats twelve broken FAQ blocks. **Validate after every deploy, and spell entity names one way everywhere.**

## Knowledge graph

A knowledge graph connects facts about a client, such as its name, locations, founders, and products, with pages that corroborate them.

**Changes.** Building that record means consistent names and sameAs links, review profiles, directory listings, and outside pages that repeat the right facts. A tool that merely displays the record measures it. The distinction matters when [third parties account for much of the citation opportunity in unbranded B2B answers](https://authoricy.com/blog/aeo-for-marketing-agencies). Before signing, look at what it takes to [manage a knowledge graph and publish to a client CMS](https://groas.com/post/agency-aeo-knowledge-graph-cms-publishing-guide). **If the platform only tracks the graph, you still own the cleanup.**

## Answer-first content

Answer-first content puts a short, self-contained answer directly below the heading that asks the question.

**Changes.** Retrieval systems can lift a passage more easily when it makes sense on its own. One [breakdown links short answer passages with featured snippets and reports roughly 2.5 times higher snippet likelihood](https://www.rankshift.ai/blog/best-answer-engine-optimization-strategies/). Vendors misuse the term when they mean _longer blog posts_. Length is not the fix. Write a direct 40–60-word answer, then expand below it. **If the block makes no sense without the rest of the page, rewrite it.**

## Passage optimization

Passage optimization shapes each section so an answer engine can lift a useful chunk without lifting the whole page.

**Changes.** Think short paragraphs, real lists, small tables, exact numbers, and no throat-clearing introduction. [Passage-level retrieval favours chunks such as short paragraphs, lists, and tables](https://www.rankshift.ai/blog/best-answer-engine-optimization-strategies/). I recognise the principle from PPC landing pages: the answer that mattered had to be easy to find, not buried beneath everything the business wanted to say. **One idea per block; answer the heading in the first sentence.**

## Direct CMS publishing

Direct CMS publishing sends approved content from the platform to the client site and records what went live.

**Changes.** No export, no ticket, no intern pasting into WordPress at midnight. Every agency owner knows the graveyard of drafts in Google Docs that never shipped because the client never sent logins. A tool that writes but cannot publish leaves you holding that queue. Ask for the connector list, approval flow, and rollback. **If publishing still means your team copy-pastes, the vendor has not taken execution off your hands.**

![Editorial illustration contrasting a page shipped with a draft stuck in a queue](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/ebf01976-6c2c-4235-baa1-ed09b570a0d5.png)

## Content gap

A content gap is a buyer question for which the client has no page that answers well enough to cite.

**Measures.** A gap report identifies the problem; it does not close it. The valuable sequence is brief written, draft produced, page shipped, prompt retested. Most vendors stop at the list. I would pay for the list if the same system closed it. **Do not buy a gap report without asking who owns publication.**

## AEO vs GEO

AEO focuses on making answers easy to extract, while GEO focuses on the authority that helps an engine decide whom to name.

**Changes.** [That distinction separates answer-first structure and clean headings from entity work and third-party corroboration](https://question.marketing/aeo/aeo-vs-geo-vs-seo). Vendors use the terms interchangeably, but the work is different. Extraction without authority leaves a clean page nobody trusts. Authority without extraction leaves a trusted brand with nothing quotable to lift. When a tool says GEO, ask which outside pages it builds. When it says AEO, ask which client pages it rewrites. **If it does neither, it measures.**

## White-label

White-label means an agency puts its name on delivery performed by another provider.

**Changes** only when that delivery changes something. Reports, dashboards, and published pages can all carry your logo, but the margin depends on who does the work. [Wholesale runs $800–$3,500 per client per month, resold at $2,000–$5,000 for mid-market at 50–65% margin](https://authoricy.com/blog/aeo-for-marketing-agencies). That margin evaporates if your team still writes every brief and chases every login. **White-label the execution first, the logo second.** A dashboard with your colours is still a dashboard.

## Done-for-you vs self-serve

Self-serve gives your team tools to act; done-for-you delivery performs the work under guardrails you approve.

**Changes** is the claim to test in a done-for-you pitch. Self-serve gives you logins, prompts, and a gap list. Your writers, your developers, your problem. Done-for-you should ship pages, schema fixes, and citation work. Say you pay $99 per domain, then spend six hours per client each month acting on alerts: the subscription is not your whole cost. This will not work for everyone. If you have spare in-house content capacity and direct CMS access on every client, self-serve tracking plus your labour can win. Otherwise, ask the vendor to show how it:

- Publishes through the CMS or an edge proxy without a client development ticket.
- Fixes crawler access, schema, and inconsistent entities.
- Ships answer-first passages that close identified gaps.
- Builds third-party corroboration and retests the prompts.

**If those steps stay with your team, price the tool as self-serve.**

## Flat-rate multi-domain pricing

Flat-rate multi-domain pricing charges a fixed amount per client domain rather than increasing the bill with every prompt pack or seat.

**Changes** only if execution is included in that amount. [groas lists SEO and AI Search from $199 a month per client domain, including content, technical fixes, citations, and white-label reporting](https://groas.com/for-agencies). Other flat rates may cover monitoring while prompt packs, user seats, or domain overages cost extra. Those [per-domain and per-seat add-ons complicate agency scaling](https://seoprofy.com/blog/aeo-tools/). **Multiply the quote by your prompts, domains, and seats before comparing prices.**

## Citation building

Citation building places consistent client facts on third-party pages that an answer engine may use as sources.

**Changes.** That can mean relevant lists, review profiles, directories, partner pages, and digital PR. It is authority work that a tracking-only resale leaves untouched. A placement is not a guaranteed AI citation, though. Ask which domains the vendor works on, what it placed there, and what happened in subsequent answer checks. **A citation chart is not citation building.**

## E-E-A-T and third-party corroboration

E-E-A-T names Experience, Expertise, Authoritativeness, and Trust, qualities a client can support with identifiable authors and corroborating information.

**Changes.** In this work, look for real author pages, accurate dates and credentials, consistent Organization data, and outside mentions that match the client’s claims. An audit PDF can identify what is missing; it cannot fix an anonymous page or create outside corroboration. **If the evidence lives only in the audit, the vendor measures.**

## Entity consistency

Entity consistency means using the same client name, locations, founders, and product names across the site, schema, and third-party profiles.

**Changes.** One page says Acme Dental Studio, another says Acme Dental, and a directory says Acme Dentistry. A human may connect them; a system may not connect them cleanly enough to cite. Vendors misuse _entity score_ when they supply a number but no fixes. The work is plain: choose canonical names, add sameAs links where appropriate, clean up duplicates, and validate the schema. **Pick one spelling and enforce it everywhere.**

## AI crawler monitoring and referral traffic

AI crawler monitoring counts bot visits, while referral tracking counts clicks back from AI answers.

**Measures.** Server logs or CDN analytics may show GPTBot, ClaudeBot, PerplexityBot, and OAI-SearchBot fetching pages; referral data may show visits from ChatGPT, Perplexity, or AI Overviews. Useful diagnostic, weak goal. Bot hits do not equal citations, and more fetches do not prove an engine is convinced. I watch them the way I watched impression share in PPC: directionally helpful, never the target. **If the win is more crawls but not more citations, keep asking questions.**

## Citation-rate guarantees

A citation-rate guarantee promises a specified level of AI citations, mentions, or share of voice by a deadline.

**Measures neither a useful action nor a result the vendor controls.** A vendor can control what it ships, how quickly it ships, and how carefully it fixes entities. It cannot control the model, the prompt mix, or competitor pages that change in the same week. I used to put forecast ranges in proposals to make AEO look like PPC. I was wrong. Sell the shipping cadence and the retest loop. Report citation movement honestly. **Retire the guarantee.**

## Frequently Asked Questions

### How can I tell whether an AEO vendor actually does the work or just tracks results?

Sort every term on the vendor page into Measures or Changes. Measurement terms like prompt tracking and visibility scores record what happens; execution terms like edge proxy, schema fixes, and CMS publishing ship the work. When a rep blurs the line, ask who clicks publish. If the answer is your team, budget for your team to do the work.

### Is prompt tracking worth buying on its own?

Only if you have a way to act on the data. Prompt tracking runs a fixed set of buyer questions through AI search tools and records whether your client appears, which tells you which prompts you lose. It does not write the paragraph that wins the next run, so buy it only with a plan to fix what it finds.

### What does a tracked prompt cap actually cost me in coverage?

The cap limits how many prompts a tool runs each month, which limits what you can see. For example, 25 prompts split across two products and three funnel stages gives you a sample, not coverage. Price the cap against prompts per client per week before you price the retainer.

### What is the difference between AI share of voice and a visibility score?

Share of voice is your mentions divided by total brand mentions, yours plus competitors, times 100, so it shows whether you are losing to someone specific. A Visibility Score is answers mentioning you divided by all answers tracked, so it shows whether you show up at all. Demand the raw counts behind both, because a score without its denominator is not useful.

### Why do I get brand mentions in AI answers but no citations or links?

Rising mentions without rising citations can point to a missing citable page: the answer may be thin, the name may be inconsistent, or there may be no clean passage to lift. Mentions name the brand; citations give readers an attributed source to click. Log both separately per query and fix the page instead of celebrating the mention chart.

### What is an edge proxy and how does it help agencies ship fixes?

An edge proxy sits between a client site and its visitors or crawlers, changing the HTML they receive without changing the CMS. It can inject answer blocks, schema, internal links, and metadata without a developer. Ask who approves the diff and how you roll it back, because a proxy is execution only when changes actually ship.

### Is publishing an llms.txt file enough to get cited by AI search engines?

No. llms.txt is a Markdown file at the site root that curates LLM-friendly material, but unlike robots.txt it controls nothing and no crawler has to read it. Publish it if you want, but do not mistake the file for the work that gets cited.

### Should I trust vendors who guarantee a citation rate or share of voice?

Be sceptical of any citation-rate guarantee. A vendor can control what it ships and how carefully it fixes entities, but it cannot control the model, the prompt mix, or competitor pages that change the same week. Look for a shipping cadence and a retest loop with honest reporting on citation movement instead.

## Pay For Results, Not For Hours

Businesses buy the outcome, agencies resell it, and groas answers for it either way.

[See If You Qualify](https://groas.typeform.com/to/xC1bQNUT)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22#webpage","url":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22","name":"The AEO Vendor Glossary: 22 Terms That Measure Results or Change Them","description":"A glossary of 22 AEO vendor terms, each tagged Measures or Changes, to help agencies tell a tracking dashboard from a platform that ships the work.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/fc3582c6-c231-4152-9fdb-ab19e4c13eec.png"},"breadcrumb":{"@id":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22#breadcrumb"},"dateModified":"2026-10-12T05:20:11.068Z","mainEntity":{"@id":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"The AEO Vendor Glossary: 22 Terms That Measure Results or Change Them","item":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22#article","url":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22","mainEntityOfPage":{"@id":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"The AEO Vendor Glossary: 22 Terms That Measure Results or Change Them","description":"A glossary of 22 AEO vendor terms, each tagged Measures or Changes, to help agencies tell a tracking dashboard from a platform that ships the work.","datePublished":"2026-10-12T05:20:11.005Z","dateModified":"2026-10-12T05:20:11.068Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/fc3582c6-c231-4152-9fdb-ab19e4c13eec.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"changes, client, aeo, vendor, work, aeo vendor, terms, measures","wordCount":2238,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"Ex Goldman Sachs and Ex Stanford Computer Science","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/the-aeo-vendor-glossary-for-agencies-22#faq","mainEntity":[{"@type":"Question","name":"How can I tell whether an AEO vendor actually does the work or just tracks results?","acceptedAnswer":{"@type":"Answer","text":"Sort every term on the vendor page into Measures or Changes. Measurement terms like prompt tracking and visibility scores record what happens; execution terms like edge proxy, schema fixes, and CMS publishing ship the work. When a rep blurs the line, ask who clicks publish. If the answer is your team, budget for your team to do the work."}},{"@type":"Question","name":"Is prompt tracking worth buying on its own?","acceptedAnswer":{"@type":"Answer","text":"Only if you have a way to act on the data. Prompt tracking runs a fixed set of buyer questions through AI search tools and records whether your client appears, which tells you which prompts you lose. It does not write the paragraph that wins the next run, so buy it only with a plan to fix what it finds."}},{"@type":"Question","name":"What does a tracked prompt cap actually cost me in coverage?","acceptedAnswer":{"@type":"Answer","text":"The cap limits how many prompts a tool runs each month, which limits what you can see. For example, 25 prompts split across two products and three funnel stages gives you a sample, not coverage. Price the cap against prompts per client per week before you price the retainer."}},{"@type":"Question","name":"What is the difference between AI share of voice and a visibility score?","acceptedAnswer":{"@type":"Answer","text":"Share of voice is your mentions divided by total brand mentions, yours plus competitors, times 100, so it shows whether you are losing to someone specific. A Visibility Score is answers mentioning you divided by all answers tracked, so it shows whether you show up at all. Demand the raw counts behind both, because a score without its denominator is not useful."}},{"@type":"Question","name":"Why do I get brand mentions in AI answers but no citations or links?","acceptedAnswer":{"@type":"Answer","text":"Rising mentions without rising citations can point to a missing citable page: the answer may be thin, the name may be inconsistent, or there may be no clean passage to lift. Mentions name the brand; citations give readers an attributed source to click. Log both separately per query and fix the page instead of celebrating the mention chart."}},{"@type":"Question","name":"What is an edge proxy and how does it help agencies ship fixes?","acceptedAnswer":{"@type":"Answer","text":"An edge proxy sits between a client site and its visitors or crawlers, changing the HTML they receive without changing the CMS. It can inject answer blocks, schema, internal links, and metadata without a developer. Ask who approves the diff and how you roll it back, because a proxy is execution only when changes actually ship."}},{"@type":"Question","name":"Is publishing an llms.txt file enough to get cited by AI search engines?","acceptedAnswer":{"@type":"Answer","text":"No. llms.txt is a Markdown file at the site root that curates LLM-friendly material, but unlike robots.txt it controls nothing and no crawler has to read it. Publish it if you want, but do not mistake the file for the work that gets cited."}},{"@type":"Question","name":"Should I trust vendors who guarantee a citation rate or share of voice?","acceptedAnswer":{"@type":"Answer","text":"Be sceptical of any citation-rate guarantee. A vendor can control what it ships and how carefully it fixes entities, but it cannot control the model, the prompt mix, or competitor pages that change the same week. Look for a shipping cadence and a retest loop with honest reporting on citation movement instead."}}]}]}
```