# Answer-engine visibility

Track whether AI assistants and AI Overviews mention, recommend and cite the client, and fix it when they stop.

A growing share of buying research never reaches a results page. LLM Visibility measures how the client appears when someone asks an AI assistant instead of searching.

Three panels: Test runs a fresh visibility check, Analytics shows the current standing across assistants, and Reports keeps the history of previous runs so you can compare over time.

## How it works

1. **Build a prompt set**: Prompts are the real questions a buyer asks: "best plumber in Austin", "who repairs tankless water heaters near me". The platform can generate a starting set from the client profile.
2. **Run the checks**: Each prompt is run and the response is analysed for brand mentions, position in the list, sentiment, and which sources were cited.
3. **Read the dashboard**: Mention rate, share of voice against named competitors, sentiment, and the domains the answer engines trust for this topic.
4. **Schedule**: Prompt sets re-run on a cadence so you see trends, not one reading.

## Reading a worked example

A prompt set of 20 questions for a local HVAC client might return a 35% mention rate, with the client appearing third when mentioned and cited less often than two competitors with more extensive, question-driven blog content. That gap, mentioned but rarely cited, usually means the client's content answers questions indirectly (through service pages) rather than directly (through a dedicated article), which is exactly the fix below.

## AI Overview monitoring

A separate daily job checks Google AI Overviews for tracked queries and records whether the client is cited. Losing a citation raises an anomaly alert, because it usually costs traffic before rankings move.

## Improving visibility

- Answer questions directly and early on the page. Answer engines lift concise, factual passages.
- Publish the specifics assistants need: service areas, pricing structure, credentials, brands serviced, hours.
- Keep structured data accurate; it is how machines confirm what a page is about.
- Earn mentions on the sources that already get cited for the topic. Those citation lists are a link-target shortlist.
- Publish an llms.txt style summary of the business so assistants have a canonical description to read.

> **Report this to clients**: Most clients have never seen whether AI assistants recommend them. Share of voice against named competitors is one of the easiest wins in a monthly report.

## Common problems

**Mention rate is low even though rankings are strong**

Strong rankings do not guarantee an LLM mention. Assistants favour pages that answer the question directly in the first few sentences. Check whether the top-ranking page actually states the answer plainly near the top.

**Client is mentioned but never cited as a source**

This usually means a competitor's content is the one actually being quoted. Look at the cited-domains list in the dashboard and see what format their content uses that the client's does not.

**Sentiment reads negative or mixed**

Check what source the assistant is likely drawing from, often an old negative review or an outdated third-party listing. Address the underlying source rather than the AI response itself.

**AI Overview citation disappeared**

Check whether the previously-cited page changed recently (content edit, template change, or a broken canonical), losing a citation often follows a page change more than an algorithm shift.