Product

Visibility Sentiment Competitors Citations AI Traffic Opportunities Pricing

Engine trackers

ChatGPT Google AI Overviews Perplexity Claude Gemini Grok Meta Spark AI Mode

Resources

How it works Blog Learn Log in
AI Sentiment

What AI thinks of you and the words it uses to say it.

Every mention is scored across the aspects that matter (pricing, quality, support, whatever fits your business) into a strengths-and-weaknesses map.

Join the waitlist

7-day trial, no card

Perception snapshot 21 Aug 06:00
Sentiment score
+38 +6
Classified mentions
147
Strongest aspect
Quality +62
Weakest aspect
Service -29
Sentiment by aspect
Quality
+62
Speed & availability
+41
Range & features
+24
Trust & reputation
+18
Price & value
-18
Service & support
-29
Your taxonomy: rename, add or regenerate anytime. 6 aspects

Strengths and weaknesses, mapped

Mentions are classified into aspects (pricing, support, reliability and more) and scored, so you see at a glance where AI sells you and where it undermines you.

The sentences behind the score

Every aspect score is backed by the actual sentences AI wrote, by engine and by model. Click through from any score to the quotes behind it.

Your taxonomy, not ours

Rename aspects, add your own, or regenerate the set. The perception matrix tracks the dimensions your customers care about.

Beyond the mention

What is AI sentiment?

Being named is the first hurdle. The second is how you are described in the same breath, because an assistant does not hand over ten links for the buyer to weigh up. It hands over a characterisation, and they take it. AI sentiment is that characterisation: the tone of the mentions you get, broken down by what they were about.

Named first and described as expensive with slow returns is not a win. Third on the list and called the durable one is a different problem to solve.

Named, and characterised 21 Aug
Are Salomon trail shoes worth the money?
Salomon shoes are durable and grip well in deep mud, though they are expensive and returns can be slow. For dry trails, cheaper options do the job.
Quality + Price - Service - 1 mention · 3 aspects

Evidence, not vibes.

Pick an aspect and read the sentences that produced its score, in the engine that wrote them. Nothing here is a summary of a summary.

AspectsNet score
Quality Strong Net +62
The Speedcross upper holds up to repeated mud and grit better than most in this price range. Positive
Reviewers consistently praise the outsole compound for keeping its bite over 500km. Positive
Some long-term owners report the toe bumper separating after a heavy season. Negative
Descriptors on this aspect: durable · grippy · well-built Open the answer →

See the score, then close the gap.

Knowing you are losing is the easy part. Referenced ranks the prompts you lose most often, shows you the answer that beat you, and names the page or review that would win it back.

See how Opportunities ranks the work.

Prompts · Scan 21 Aug Answers lost ↓
Prompt GPTPPLXGEM Score
"Best trail shoes for a first ultra?" — — — 0.0
"Cheapest waterproof trail shoes?" #2 #3 — 51.0
"Best trail running shoes for muddy ultras?" #2 #1 #2 88.0

Read differently by every engine.

The same brand reads differently depending which model is doing the talking. Scores and quotes are kept per engine as well as blended overall, so you can tell a universal weakness from one model's house style.

Salomon · Sentiment by engine · 30 days
ChatGPT
+64 Strong
Perplexity
+51 Strong
Google AI Overviews
+22 Mixed
Claude
+18 Mixed
Gemini
+9 Mixed
Grok
-6 Mixed
Meta Spark
+27 Positive
AI Mode
+33 Positive

Sentiment questions, answered.

How sentiment is classified, and what to do with it.

How is sentiment scored? +

Every mention is classified positive, neutral or negative against the aspect it's about, from the exact sentences AI wrote, not a summary of the answer as a whole. Each aspect's score is the net of its positive and negative classified mentions.

Is negative sentiment with high visibility bad? +

Being named often but described badly is worse than not being named at all. A confident wrong answer is what buyers act on. High visibility paired with negative sentiment on a core aspect like pricing or support is the clearest signal to act, not a vanity metric to ignore.

Can I edit the aspects? +

Yes: rename aspects, add your own, or regenerate the set from scratch. The perception matrix tracks the dimensions your customers care about, not a fixed list you're stuck with.

Is sentiment broken down by engine? +

Yes. Each engine can characterise you differently, so aspect scores and quotes are kept per engine as well as blended overall, useful for telling a universal weakness from a single model's house style.

What exactly is the net score? +

The share of classified mentions that were positive, minus the share that were negative, so it runs from −100 to +100. Neutral mentions count towards the sample but pull the score neither way.

Why does an aspect say "low sample"? +

Nothing is scored on fewer than five classified mentions. Below that a single sentence would swing the number, so the aspect reports its count and waits for the next scans rather than show a score you can't trust.

What are descriptors? +

The actual words an engine used about you ("durable", "pricey", "slow") lifted from the sentence and kept with their tone. They show what a score is made of, and which words you'd want to change.

Do competitors get sentiment too? +

Yes. Every brand named in an answer is classified the same way, so the perception matrix puts your aspects beside theirs and shows the gap on each one.

Is sentiment different when someone asks about me by name? +

It can be, and it's kept separately. Prompts that name you are scored on their own, so you can tell how AI describes you to someone who already knows you from how it describes you to someone who doesn't.

How often is sentiment updated? +

On every scan. Each new answer is classified as it comes in, so the score moves as the answers move, daily on the current schedule.

See your visibility score before your competitors see theirs.

Free 7-day trial, no card: the first scan runs on all prompts across three engines.

Feature

Visibility

Your score across every AI engine.

Feature

Sentiment

What AI thinks of your brand.

Feature

Competitors

Share of voice, brand by brand.

Feature

Citations

Which pages AI actually cites.

Feature

AI Traffic

GA4 + Search Console, overlaid.

Feature

Opportunities

The fix for every gap, ranked.