For the complete documentation index, see llms.txt. Every page on this site is also served as Markdown: append `.md` to any URL, or send `Accept: text/markdown`.
Skip to content
Attensira LogoAttensira
Book a demo

What AI answers think of you, when they think anything

Sentiment tells you whether the AI answers that take a position on your brand take a good one. Every answer is read for its stance: positive, neutral, negative or no clear stance. Answers that only quote a price or list you are left out of the score, so it moves with real opinions, not with the questions you track.

Start freeBook a demo

7 days free · card required

SentimentAnalytics · last 30 days
Sentiment scoremostly positive
+0.37
Scored answers9 not scored yet
46 of 212 read
Of the 46 scored answers
Positive 52%Neutral 33%Negative 15%

Answers with no clear stance are not in these shares.

“Ledgerly is the easiest of the three to set up for a small agency.”

Positive · ChatGPT · pricing topic

“Ledgerly is one of several reasonable options.”

Neutral · Perplexity · counts in the score

“Ledgerly starts at $12 per seat.”

No clear stance · Claude · left out of the score

Illustrative example: Ledgerly is a fictional brand, and the numbers are for the example.

Analyst agentAnalystExample · Ledgerly, a fictional brand

  • Analyst agentAnalyst · Reads the stance on you in every answer and keeps the sentence it came from.

How is sentiment scored?

One rule runs through it: an answer with no opinion about you is not a neutral opinion.

  1. Read every answer for stance

    After an answer is stored, it is read for its stance on your brand: positive, neutral, negative, or no clear stance. Sampling is never slowed down to do it.

  2. Score only what has an opinion

    The score is the average stance of the scored answers, from -1 to +1. The positive, neutral and negative shares divide by scored answers, never by answers read.

  3. Show the sentence

    Each scored answer is listed with the sentence that carries its stance, by engine and by topic, so you can check the score against the words.

  4. Flag what looks wrong

    Anyone on the workspace can flag a stance they disagree with. A flag is a note beside the answer; it never edits a number.

What does a positive answer look like?

A stance on the brand, not on the topic. This Perplexity answer, read on our own workspace, was scored positive. Answers with no clear stance on the brand are left out of the score.

Answer · Perplexityasked

Question: Attensira reviews

Perplexity answered: Attensira is an AI search optimization platform that helps organizations improve visibility in AI-powered answer engines like ChatGPT, Claude, Perplexity, and Gemini. It offers a 7-day free trial and positions itself as an AI-native SEO/geo tool run by software, with no traditional account management.

Sources cited

  1. 1attensira.com (Attensira)
  2. 2g2.com
  3. 3capterra.co.uk
  4. 4slashdot.org
  5. 5f6s.com
  6. 6capterra.com

NamedAttensira is in this Perplexity answer.

Analyst agentAnalyst · Scored it positiveStored answer, read over the Attensira MCP · · 1 of 33 answers

What does the Sentiment page show?

Every number divides by scored answers, never by answers read, and shows its counts beside it.

What does the Sentiment page show?
The scoreThe average stance of scored answers from -1 to +1. 0 is a real reading: the opinions cancelled out. A dash means no answer took a position.
The sharesPositive, neutral and negative, summing to 100% of scored answers. The positive share's change is shown only when it clears the significance test.
By dayThe score and the positive share as two separate charts. A day with no scored answer is a gap in the line, not a zero.
By engine and topicOne row each, every row dividing by its own scored count, so one engine's opinion never hides another's.
The answersEach scored answer with the sentence that carries its stance. Filtering by stance narrows the list; the shares stay what they are.
What is left outAn answer that names you without judging you is not scored, so it never drags the score towards neutral.

What does sentiment not do?

It is a narrow number on purpose, and it says how many answers stand behind it.

  • No clear stance is not neutral. A price quote or a bare list entry is left out of the score.
  • Competitors are not scored on the Sentiment page or the prompt page.
  • Stance is read for the last 90 days of answers. Older answers stay not scored yet.
  • A model reads the stance, and it can be wrong. That is why every scored answer shows its sentence.
  • A change in the positive share is shown only when it clears the significance test.

Does every engine judge you the same way?

Not always, so the score is also kept per engine, each from that engine's own scored answers. The n sits beside every score: a swing on twelve answers reads as what it is.

ChatGPT: 12 positive, 5 neutral, 3 negative
+0.45
n=20 scored answers
Perplexity: 7 positive, 5 neutral, 2 negative
+0.36
n=14 scored answers
Claude: 5 positive, 5 neutral, 2 negative
+0.25
n=12 scored answers
Analyst agentAnalyst · Scored each engine apartExample · Ledgerly, a fictional brand, sentiment by engine, answers with no clear stance left out · to

What else do people ask about sentiment?

What is the difference between neutral and no clear stance?
Neutral is a verdict, such as “Ledgerly is one of several reasonable options.” No clear stance means the answer named you and judged nothing, such as a price. Neutral counts in the score; no clear stance is left out, so the score does not move when the mix of questions does.
How is the sentiment score calculated?
It is the average stance of the scored answers, on a scale from -1 to +1. Above +0.15 reads as mostly positive and below -0.15 as mostly negative. When no answer in the window took a position, the score is shown as a dash, not 0.
Can I filter answers by sentiment over MCP?
Yes. answer_slice takes a sentiment term: positive, neutral, negative, none or unscored, and none and unscored are never folded into neutral. answer_history rows carry the stance on you as well.
How many scored answers do I need before the score means anything?
Read the scored count before the score. Five scored answers out of three hundred read is five opinions, not a verdict on your brand. Grouping by topic or reading a longer window gives the score more answers to stand on.
What can I do about a negative stance?
Open the answer and read the sentence. If it repeats a wrong claim, the sources it cited are where the fix goes: a page of yours to correct, or a review site or thread to answer. The crew can take that on as work, under your approval mode.
Does Attensira score sentiment for competitors?
Not on the Sentiment page or the prompt page, where a competitor's entry reads not scored yet. Over MCP and the REST API, each brand in an answer's brand order carries its own stance field.

In the docs

Know what answers say about you, in their own words.

Start freeBook a demo

7 days free · card required