> ## Documentation Index
> Fetch the complete documentation index at: https://docs.growthmention.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Sentiment and stance
> Whether AI assistants recommend your brand, list it, or warn against it — and the tone they use.
Being named isn't the same as being recommended. For every brand in every answer, Growth Mention records two judgements: its **stance** — what role it plays in the answer — and its **sentiment** — the tone used about it. Both come with the sentence from the answer that justifies them, quoted exactly.
## Stance
Stance is the more useful of the two. Answers to *"what's the best…"* questions are warm about almost everyone they name, so tone barely separates brands — but the answers are very clear about which ones they *recommend* and which they only *list*.
| Stance | What the answer does with the brand | Example |
| --------------- | ---------------------------------------------------- | ---------------------------------------------------- |
| **Recommended** | Puts it forward as a good choice. | *"For a small team, I'd go with Acme."* |
| **Conditional** | Recommends it, with a real caveat. | *"Acme is solid, but expensive for under 10 users."* |
| **Listed** | Names it as an option, with no verdict. | *"Popular options include Acme, Beta and Gamma."* |
| **Cautioned** | Warns against it. | *"Users report Acme's support is slow."* |
| **Comparison** | Uses it as a yardstick, not as an option. | *"A cheaper alternative to Acme."* |
| **Context** | Mentions it in passing — an integration, a platform. | *"Integrates with Acme and Beta."* |
Only the answer's own words count. A brand's reputation elsewhere never affects its stance in an answer.
## Sentiment
| Sentiment | Meaning |
| ------------ | ------------------------------------------- |
| **Positive** | Praise, a named strength, a recommendation. |
| **Neutral** | No evaluation — the brand is just named. |
| **Negative** | A weakness, a complaint, a warning. |
## The scores on the Sentiment page
**Recommendation rate** is the share of your mentions that are *recommended* or *conditional*. It's the clearest single measure of whether the assistants are sending people your way.
**Sentiment index** turns tone into one number from 0 to 100: each positive mention counts 1, neutral ½, negative 0, averaged. **50 means neutral.** Because recommendation questions draw mostly positive answers, indexes in the 80s and 90s are common for the brands they name — so compare yours with your competitors' rather than with 100, and watch for drops.
When brands are ranked by recommendation rate, small samples are adjusted for. A brand recommended 7 times out of 7 doesn't rank above one recommended 53 times out of 60 — seven answers are too few to be sure.
**Click a brand** in the standings to see what was said about it: every judged sentence, with its stance, its tone, the assistant, the day and the prompt it answered. Anything critical comes first. Click **All brands** to go back.
**Sentiment by model** shows each brand's index on each assistant, so you can see where you're treated worse than elsewhere.
## Where to act
The Sentiment page lists **prompts where a competitor is recommended and you're only listed, or not named at all** — worst first. Those are the questions where the assistants actively send buyers elsewhere, and the evidence quotes show you exactly what they say.
