Sentiment

The Sentiment page shows how AI describes your brand — positive, neutral or negative — in the answers that mention it.

For the basics of how sentiment is calculated, see What SimplyRank measures.

What you see

1. Five headline numbers

Net sentiment — the share of positive mentions minus the share of negative ones, so it runs from −100 (all negative) to +100 (all positive). It shows green above +25 and red below −10. The line underneath says which tone leads, for example "Mostly positive".
Positive, Neutral and Negative — each tone's share of the scored mentions, with how many mentions that is.
Mentions scored — how many mentions the numbers rest on, and how many days in the window had any.

Under the tiles, one bar splits every scored mention into its three tones:

Positive — AI describes you favourably ("a leader in...", "well-regarded for...").
Neutral — AI mentions you factually without endorsement ("options include..., among them...").
Negative — AI describes you unfavourably ("users complain that..., concerns include...").

A healthy brand sits in the 70–100% positive band described in What SimplyRank measures, with neutrals making up most of the remainder and negatives in single digits. Departures from that need attention.

Only answers that mention your brand are scored, and only scored answers are counted. An answer that leaves you out has no tone to measure; Gap analysis covers those.

2. By model

One row per AI model, the model with the most mentions first. Each row has a bar of its three tones with their shares, its net sentiment on a coloured chip, and how many mentions it rests on.

When one model stands well apart from the others — far more negative, or far more neutral — a line under the table names it. It only does so when there are enough mentions on both sides for the difference to mean something.

If you filter the page to one model, the table shows just that model's row.

3. Responses

The answers behind the numbers, one tone at a time. Tabs for Negative, Neutral and Positive show how many mentions of each tone were scored; the list opens on the first tone that has any, negative first.

Each row shows:

Prompt — the question asked, with its category and intent.
Model — which AI answered, marked Sentiment probe when the answer came from one.
Score — how favourably the answer describes you, from 0 to 100.
Why it was scored this way — the reason recorded when the answer was scored.
Seen — when the answer was captured.

Click a row to read the part of the answer around your brand's name, with the name highlighted. View full answer opens the whole answer.

The list shows twenty answers at a time, newest first; page through with the Previous and Next buttons under it. The search box looks through the answers on the page you're on only.

The list finds answers by your brand's name over the chosen number of days up to now, and tags don't narrow it, so its counts can differ from the numbers above. The coverage note says so too.

4. Sentiment probes note

If your workspace has no sentiment probes yet, a note at the foot of the page says so, with a Create sentiment prompts link. Sentiment probes ask the models directly how they rate you, alongside the tone of the answers to your visibility prompts.

Filters and controls

The controls sit beside the page title:

Coverage note — when the data was captured and how many days had scored mentions. Open it for the window, where the scored mentions came from and how each number is worked out. Its dot turns amber when fewer than five mentions were scored, because numbers that small can swing a lot from one scan to the next.
Filters — Model, Category, Intent, Source and Tags. They apply to every card, though tags don't narrow the responses list. Source lets you look at visibility scans or sentiment probes on their own: the two ask different questions, so comparing them is worth doing.
Date window — from 7 to 365 days. The default is the last 30 days.

The Sentiment page has no CSV export.

How to act on negative sentiment

A few moves that work, in order of leverage:

1. Read the negative responses first

Open the Negative tab and read the reasons and the answers themselves. Patterns you'll find:

A specific old review or news story that keeps coming up.
A specific complaint thread on Reddit.
A factually incorrect piece of content the AI is treating as authoritative.

To see which sources AI is citing, open Citations. If AI keeps leaning on one negative review or article, engage with that source:

Respond to the review publicly (most platforms allow it).
Reach out to the author of an outdated article and ask for a correction or update.
Earn fresher, positive citations to dilute the old ones.

2. Improve the volume of positive third-party signals

Sentiment is a ratio, so the fastest lever is usually the denominator. Suppose AI currently sees 2 positive mentions and 3 negative ones — that's 40% positive, and adding 5 more positive mentions moves you to 70% without a single negative one going away. Tactics: refresh G2/Trustpilot listings, ask happy customers for reviews, pitch a positive feature story.

3. Audit your own site

Sometimes the negative signal originates on your own domain — old crisis communications, deprecated product pages, unflattering case studies. Audit, update, and remove.

4. Don't argue with AI

You can't directly tell AI to update its picture of you. The path is changing the underlying signals. Patience and consistency win.

Plan availability

The Sentiment page is on every plan, including the free trial.