What is brand sentiment?
Brand sentiment is the balance of positive, negative and neutral opinion expressed about a brand across a defined set of sources over a defined window. It answers how people talk about you, where mention volume answers how much. Most teams report it as a net sentiment score: positives minus negatives, divided by total mentions. There is no standard formula and no standard source list, so a sentiment score is only comparable with another one computed the same way. This page covers how the metric is built, why it moves, and what it does not measure.
Brand sentiment, defined
The definition carries three conditions that do most of the work, and dropping any of them makes the number unreadable. A defined source set, because sentiment measured across review platforms and sentiment measured across social media are different quantities. A defined window, because a rolling thirty days and a rolling year respond to events at different speeds. And a defined classification method, because the labels come from software or from people, and switching between them moves the result on its own.
Sentiment, volume and reputation are three different things
These get used interchangeably in reporting and they answer separate questions.
| Measure | The question it answers | What it misses |
|---|---|---|
| Mention volume | How much is being said about us | Whether any of it is favourable |
| Share of voice | How much of the category conversation is ours | Whether the conversation is one we want |
| Brand sentiment | How favourable is what is being said | Who is not saying anything, and why |
| Reputation | What people expect of us, and whether they act on it | Almost everything a dashboard can see |
Reputation is the expectation that governs whether someone hires you, refers you or believes you when something goes wrong. Sentiment is one observable trace of it, drawn from the minority of people who bothered to write something down. Treat sentiment as an indicator with a known sampling problem and it is genuinely useful. Treat it as reputation itself and you will manage the indicator instead of the business.
How brand sentiment is measured
The recipe is short and the discipline is in keeping it identical between periods.
- Fix the sources. List them explicitly: which review platforms, which social networks, which forums, whether news is included, whether employee sites are included.
- Fix the window. Rolling thirty days is the common default for operational reporting; calendar months are easier to explain to a board.
- Collect the mentions. Decide up front how you handle mentions that name you without your exact brand string, and how you exclude the ones that are a different company with a similar name.
- Classify. Automatically, by hand, or both. The method behind the labels, and its failure modes, are covered in sentiment analysis.
- Compute the score. See the formula below, and write down which one you used.
- Break it down by topic. This is the step most teams skip and the one that makes the number actionable.
- Chart it against events. Launches, price changes, outages, campaigns, staffing changes.
The sentiment score, and why there is no standard one
The most common formula is net sentiment: positive mentions minus negative mentions, divided by total mentions, expressed as a percentage. On that formula the scale runs from negative one hundred to positive one hundred, and neutral mentions pull the result toward zero by inflating the denominator.
A second common formula divides positives by the sum of positives and negatives, ignoring neutrals entirely, and produces a number from zero to one hundred. It is not wrong, it is a different question, and it will usually look better. Neither formula is standard, which means a sentiment score is only comparable to another sentiment score computed the same way, from the same sources, over the same window, by the same classifier.
The practical rule follows from that: write your formula down, put it in the footer of every report, and never change it quietly. If you must change it, restate the previous periods on the new method so the trend line stays honest.
Why the number moves when nothing has changed
Before treating a movement as a result, rule out the boring explanations.
- A source was added or dropped, or a platform changed what its API returns.
- The classifier changed. A vendor model version update reweights every label, retroactively in some tools.
- One item went wide. A single thread with hundreds of replies can swamp a month for a small brand.
- Seasonality. Complaint volume in a service business tracks demand.
- Volume fell. Net sentiment computed over forty mentions is noise wearing a decimal point.
- Someone ran a campaign. Asking every customer for a review, which is entirely legitimate, changes the mix by design.
What brand sentiment is good for, and what it is not
It is good for four things: spotting a change early, comparing the same measure across periods, ranking which topics generate the most negative volume, and giving a communications team something concrete to report other than impressions.
It is poor at four others. It does not attribute cause, so a fall tells you to go looking rather than what to fix. It does not predict revenue, and any model claiming it does is fitting a curve to a small sample. It is not proof of damage in any legal sense: harm to reputation is argued with evidence about specific statements and specific consequences, and the reason most negative material stays online at all is structural, set out in the text of Section 230 at Cornell's Legal Information Institute, which makes platforms not the publisher of what their users post. And it is not comparable to another company's reported sentiment, because you cannot see their method.
It also is not the metric that moves your visibility. Google's guidance on local ranking factors names review count and review score among the inputs to prominence in local results. Your sentiment dashboard is a management instrument. Your star rating is an input to whether people find you at all.
Sentiment and the single number problem
Compressing many opinions into one figure invites the same misuse every headline metric attracts, which is the substance of the long running argument about Net Promoter Score. The Harvard Business Review article that introduced Net Promoter Score made a case for one number precisely because a single figure travels through an organisation. That is the strength and the flaw together: it travels, and the caveats do not travel with it.
The defence is to make the caveat part of the number. Never report a sentiment score without the mention count beside it, the window it covers, and the top three topics driving the negatives. A score of positive 41 means nothing. Positive 41 across 612 mentions in July, with negatives concentrated in delivery timing, is a briefing.
Reporting cadence
Weekly for the team that acts on it, with the topic breakdown. Monthly for leadership, charted against the previous twelve months so a single bad week does not read as a trend. Real time for alerts only, and only on named triggers, because a sentiment feed watched continuously produces reaction rather than management.
Set the baseline before you start changing anything, and date it. Establishing that baseline across reviews, mentions and search results at the same time is one of the things a reputation audit is for.
Questions about what is brand sentiment?
What is brand sentiment?
The balance of positive, negative and neutral opinion expressed about a brand across a defined set of sources over a defined window. It describes how people talk about you rather than how many people do.
How do you measure brand sentiment?
Fix your sources and your window, collect the mentions, classify each as positive, negative or neutral, then compute a net score and break it down by topic. Keeping the method identical between periods matters more than which method you pick.
What is a good brand sentiment score?
There is no benchmark worth quoting, because no two organisations compute the score from the same sources with the same classifier. The only meaningful comparison is your own score against your own previous score on an unchanged method.
What is the difference between brand sentiment and sentiment analysis?
Brand sentiment is the business metric you report. Sentiment analysis is the software method that produces the labels the metric is built from. The metric inherits every limitation of the method.
Is brand sentiment the same as reputation?
No. Reputation is the expectation people hold and act on. Sentiment is one visible trace of it, sampled from the minority of people who wrote something in public.