Sentiment Score
Quick answer
A sentiment score is a numeric or categorical rating, such as positive, neutral or negative, assigned to a piece of text, like a mention, review or comment, to summarize the tone of what was said about a brand without requiring a human to read every single one.
Why it matters
A brand getting dozens of mentions a week can't have a person read and judge each one, so a consistent sentiment score is what makes it possible to spot a shift, such as a sudden run of negative mentions, without manually reading the full volume. It turns an unmanageable flood of text into a number a team can watch and act on, with the underlying text still available when the number demands a closer look. A score is also a way to compare across brands. Applying the same scoring method to a brand and its named competitors shows who is discussed more favorably in the same period, which a pile of unscored mentions cannot.
How to measure it
Apply a consistent sentiment classification to every mention collected across communities, news, blogs, social and AI answers, track the overall mix of positive, neutral and negative over a fixed rolling window, and spot-check a sample of the scored mentions periodically against human judgment, since automated sentiment scoring makes real mistakes, especially with sarcasm or niche jargon the model wasn't trained to catch. Report the mix of positive, neutral and negative mentions alongside the volume, not a single average, because an average can stay flat while the underlying mix shifts. Keep the scoring method fixed between periods, and when it changes, note the date so a trend is not mistaken for a method change.
Example
A support team notices overall sentiment dipped for a week and assumes a widespread problem. Reading a sample of the negatively scored mentions shows most of them are the same sarcastic joke about a minor UI change being repeated across a few threads, not a real reputational issue, a distinction only visible by checking the underlying text behind the score rather than reacting to the number alone.
Frequently asked questions
Is sentiment scoring always accurate?
No. Automated sentiment scoring makes real mistakes, particularly with sarcasm, slang or niche jargon, which is why spot-checking a sample against human judgment matters more than trusting the score blindly.
What's a good sentiment score?
There's no universal good score; what matters is the trend for a specific brand over time and how its mix compares to named competitors over the same window.
Does sentiment scoring apply to AI-answer mentions?
Yes. A mention inside an AI engine's answer can be scored for tone the same way a community post or news article can.
How often should a team spot-check its sentiment scores?
Periodically rather than constantly; a quick sample review once a month or so is usually enough to catch systematic scoring mistakes before they skew a trend.
Related terms
Related
MarketHQ tracks brand mentions across communities, news, blogs, social and AI answers, and turns them into gap analysis and action plans.