Guide

Brand sentiment analysis: how to measure what people feel about your brand

Brand Mentions team · November 10, 2026 · 4 min read

Short answer: Brand sentiment analysis means sorting what people say about your brand into positive, neutral and negative, and tracking how that mix changes over time and by topic. You collect mentions (web, news, reviews, social), label each one, by hand or with an AI model, and look at the share of negative mentions and the themes behind them. The number alone is not the point: the themes behind negative mentions are your to-do list.

Why it matters

  • Early warning: a rising share of negative mentions often shows up before churn or sales drops.
  • Prioritising replies: negative mentions in places customers read deserve a fast answer.
  • Measuring changes: did the new support process or pricing change shift what people say?
  • Competitive context: your sentiment means more next to your competitors'.

How automated sentiment works (in plain terms)

Older tools used word lists: "great" counts as positive, "terrible" as negative. Modern tools use language models that read the whole sentence, so they handle phrases like "not bad at all" better. Either way, the tool reads text and outputs a label (and sometimes a score).

Where it still fails:

  • Sarcasm and irony: "Great, another update that breaks everything."
  • Mixed opinions: "Love the product, hate the support."
  • Brand vs topic: a negative article about an industry problem that mentions you neutrally.
  • Homonyms: results about another company with your name.
  • Short snippets: a title alone rarely tells the full story.

That's why sentiment works best for sorting and trends, with a human reading the important ones.

A simple monthly method

  1. Collect the month's mentions from your sources: a web and news digest such as Brand Mentions, your review sites, your main social network.
  2. Label each as positive, neutral or negative (automated labels are fine; check the negatives yourself).
  3. Tag the topic of each negative and positive mention: price, quality, support, delivery, ease of use.
  4. Calculate the share negative: negatives ÷ all mentions, and the top three topics behind them.
  5. Compare with last month and, if you can, with one competitor.
  6. Decide one action for the top negative topic, and one way to amplify the top positive one.

Example

Month Mentions Negative Share negative Top negative topic
September 40 6 15% Delivery times
October 52 4 8% Pricing confusion

What to do: the delivery fix worked; now clarify the pricing page. (Example figures for illustration.)

Look at sentiment by source

The same brand can look very different depending on where you look, so split your monthly numbers by source:

  • Review sites skew to experiences: delivery, support, quality. They're closest to what customers feel after buying.
  • News and blogs skew to announcements and comparisons: mostly neutral, with occasional strong opinions in reviews and roundups.
  • Forums and communities hold the most detailed complaints and recommendations, often answering "should I buy this?"
  • Social media reacts fastest and loudest, including to things that have nothing to do with your product.

A rising share of negatives on review sites usually deserves more attention than a noisy week on social media. If you can only track one source well, track the one your customers read right before they buy.

Net sentiment, and why not to obsess over it

Some teams report a single "net sentiment" figure (positive minus negative, as a share of all mentions). It's handy on a slide but hides the useful part, which is why. Always report the themes next to the number.

Turning sentiment into decisions

  • Product: recurring negatives about a feature go to the roadmap discussion with real quotes.
  • Support: spikes in "no reply" complaints point to staffing or process issues.
  • Marketing: recurring positives tell you which benefits to lead with.
  • PR: negative coverage in an outlet your customers read deserves a response or a correction request.

Common mistakes

  • Treating neutral as good news. Neutral often means "listed but not recommended".
  • Comparing months with very different volumes without noting it.
  • Letting the tool decide replies. Read before you respond.
  • Ignoring where the mention appears. A complaint on a major review site outweighs a scraper site.
  • Analysing without acting. End every review with one decision.

FAQ

Is automated sentiment accurate enough? For sorting and trends, usually yes; for individual decisions, read the mention yourself.

How many mentions do I need? Trends become meaningful with a few dozen mentions a month; below that, read them all and skip the percentages.

Where do I get mentions from? Free alerts (see how to set up Google Alerts), review sites, and a deduped web and news digest. Related: what social listening is.

Sources