<!-- The markdown copy of https://oppalerts.com/docs/app/visibility/llm_top_sites . All pages: https://oppalerts.com/docs/llms.txt -->

[Documentation](https://oppalerts.com/docs/) / [App](https://oppalerts.com/docs/app/) / Competitor LLM Recs

# Competitor LLM Recs

`App https://oppalerts.com/app/visibility/llm_top_sites`

## What you use it for

**Which sites the AI models name** when they are asked about your market. It is the closest thing this application has to "who does AI think the market is".

- **Find where you are missing.** A site every model names, that you are not on, is the gap.
- **Build a target list.** Every site here is a site AI tools already associate with your market.
- **See how widely a site is named.** The Models column counts how many different models named it.
- **Show a client the market as an AI sees it**, which is usually not the market as they see it.

## How it works

![The table.](https://oppalerts.com/img/docs/app/visibility/llm_top_sites/the-table.png)

*The table.*

During campaign research, every model was asked which sites it would recommend. This page shows those answers.

**The score is a position score, not a count.** Each time a site is named it earns the weight of the model that named it, divided by its position in that answer to the power 1.5. Being named first is worth much more than being named tenth. It is worked out when the page loads and is never stored, so changing a model's weight changes this page at once.

**Seen is the raw count** of how many times the site was named across every answer. Score and Seen disagree often, and that is the point: a site named tenth in every answer has a high Seen and a low score.

Sites merge on their host name, lowercased and with `www` removed, so one site is one row.

**This is what the models say, not what a search engine returns.**

## On your own campaign

### The order to work in

1. **Read the top 20.** That is the market's consensus.
2. **Find your own site** and note its position. Being absent entirely is a different problem from position 30.
3. **Tag the sites that matter** on the Settings tab of Edit Campaign. Every report labels them from then on.
4. **Take the publishers to Link Prospecting** and the untagged domains to Opportunity Types.

### Reading your own position

- **Absent from every model.** The models do not associate you with this market at all. That is a presence problem, and backlinks and third-party coverage are what move it.
- **Named by one model only.** The Models column shows 1. A single model naming you often learned it from a single strong source.
- **Named by all, always low.** You are in the market and not the default answer. Look at what the sites above you have that you do not.

### Reading somebody else's position

A site at the top that is not a competitor is usually a publisher, a directory, or a review site. **Those are the pages to be on**, because they are what the models learned the market from. Take them to Opportunity Types and find out what they accept.

### When it looks wrong

- **Sites from the wrong market.** The campaign's Industry is wrong. The question the models are asked names that Industry.
- **Almost no rows.** The collection did not finish.

### Score against Seen

**Score** weighs where a site appeared. **Seen** counts how many times. They disagree often, and the disagreement is informative: a site named tenth in every answer has a high Seen and a low score, and it is a site the models know about and do not lead with.

### Merging

Sites merge on their host name, lowercased and with `www` removed, so one site is one row.

### What it is not

**This is not a search ranking and it is not what the tools cite today.** These are recommendations from the models based on their training. The question is asked without web search.

## MCP

These MCP tools give the same data. Each page has every parameter and an example.

- [`llm_answers_top_sites`](https://oppalerts.com/docs/mcp/llm_answers_top_sites): The domains the AI models name when they are asked which sites to use in this market, scored across every model and run.

## API

These API routes give the same data.

- [`POST /v1/llm_answers/top_sites`](https://oppalerts.com/docs/api/llm_answers_top_sites)

## Related pages

- [Competitor Mentions](https://oppalerts.com/docs/app/visibility/competitor_mentions)
- [Competitor Links](https://oppalerts.com/docs/app/visibility/competitor_links)
- [All App Pages](https://oppalerts.com/docs/app/)
