How to Measure Brand Sentiment Across Social Media

TL;DR

Start with a precise brand profile, collect relevant conversations over a defined lookback window, classify sentiment toward the brand, group the drivers, and manually review uncertain or high-impact results.

Define the subject before measuring sentiment

A reliable analysis begins with the brand, products, competitors, aliases, markets, and channels in scope. This context helps distinguish a genuine brand discussion from a coincidental keyword match.

Use a repeatable measurement workflow

• Choose the platforms and lookback period before collecting data.
• Filter for contextual relevance before assigning sentiment.
• Measure sentiment toward the brand, not the general tone of the content.
• Group recurring reasons into sentiment drivers and track their movement.
• Review ambiguous, sarcastic, multilingual, and mixed-sentiment examples manually.

Compare like with like

Platform mix and time window can change the apparent result. Trend comparisons should use consistent sources, definitions, and periods, and reports should make those boundaries visible to the reader.

Calculate the core metrics

• Net sentiment = positive percentage minus negative percentage.
• Negative-sentiment recall = correctly identified negative items divided by all human-labelled negative items.
• False-negative rate = missed negative items divided by all human-labelled negative items.
• Human agreement = items where reviewers agree divided by items reviewed; define whether agreement is pairwise or adjudicated.
• Manual-review rate = items sent for review divided by all analysed items.

Use a repeatable report structure

Record the brand definition, aliases, sources, lookback window, collected and excluded counts, positive/neutral/negative distribution, net sentiment, top drivers, changes versus the previous equivalent period, uncertain cases, and source-linked evidence. Keep platform-level results available because one high-volume source can dominate an overall score.

See an example brand sentiment report →Review ISO guidance for measurement uncertainty →

Method limitations

A measurement is only comparable when collection, relevance rules, languages, sources, and class definitions remain stable. Document model or prompt changes, sample manually across all sentiment classes, and avoid presenting unreviewed automated labels as ground truth.

Try MySocialAntenna

Understand what people are saying about your brand, and why.

MySocialAntenna analyses brand conversations across Reddit, X, LinkedIn and YouTube, with sentiment drivers and evidence linked to the original source.

PB

Author

Priyanka Basavalingaiah

Priyanka Basavalingaiah is a product leader and former software engineer with 15 years of experience building technology and data products. She is the founder of MySocialAntenna, where she builds buyer-intent discovery and evidence-backed social intelligence systems.