Counting mentions tells you how much people are talking about you. Sentiment tells you how they feel. A spike in mentions could be great news or a brewing crisis. Sentiment analysis adds that missing context and turns volume into meaning.
How sentiment is measured
| Method | How it works | Strengths | Weaknesses |
|---|---|---|---|
| Manual coding | People read and label mentions | Accurate, catches nuance | Slow, limited volume |
| Rule-based tools | Keywords and rules assign tone | Fast, transparent | Misses context and sarcasm |
| Machine learning tools | Models trained on labelled text | Handles volume and some context | Errors with slang, sarcasm and mixed views |
| Hybrid | Tools label, people check samples | Balance of speed and accuracy | Needs time for checking |
The Academy's glossary explains sentiment.
Common challenges
- Sarcasm: great, another delay reads as positive to many tools.
- Mixed views: love the food, hate the wait.
- Slang and emojis, which change meaning by context.
- Neutral mentions that are actually questions or news.
- Industry language, such as sick meaning excellent.
Doing it manually
For small businesses, manual sentiment tracking is often enough. Each week, read mentions, comments and reviews and label each as positive, negative or neutral. Note the reason, such as service, price or quality. A simple spreadsheet tallies results by month. Manual coding takes perhaps half an hour a week for a typical local business, and the reading itself often sparks ideas for content and service improvements that a score alone would never reveal.
Using tools
Social listening tools often include automated sentiment. Check accuracy by sampling: read fifty mentions and compare your labels with the tool's. If accuracy is poor, adjust settings or rely on manual checks for important decisions. The Academy's guide to social listening covers tools.
Net sentiment
Some teams report net sentiment: positive mentions minus negative mentions, divided by total mentions with tone. It gives a single trend line. Keep the raw counts too, because a stable net score can hide rising volume on both sides.
Reading sentiment trends
| Pattern | Possible meaning |
|---|---|
| Gradual rise in positive | Content, service or product improvements landing |
| Sudden negative spike | An issue, complaint or controversy |
| Rise in neutral | More questions or news coverage |
| Negative tied to one topic | A specific fixable problem |
Sentiment by topic
Breaking sentiment down by topic is more useful than a single score. You might find that sentiment about products is strongly positive while sentiment about delivery is negative. That points to a specific fix rather than a vague brand problem. Share topic-level findings with the people responsible, such as the warehouse team for delivery or the front-of-house team for service, so they can see customer feedback directly.
Sentiment during campaigns
Track sentiment before, during and after campaigns. Look at comments on campaign posts and wider mentions. A campaign that drives reach but negative sentiment may need adjusting. Platform reach depends on signals such as Instagram's documented watch time and completion, likes, shares (sends per reach), saves, relationship history✓, but reach without goodwill can backfire.
Responding to negative sentiment
Negative sentiment is information, not just a threat. Reply helpfully to individual complaints, fix recurring issues and share improvements publicly. The Academy's guide on how to respond to negative comments explains replies.
Reporting sentiment
Report the trend, the main drivers of positive and negative sentiment, notable examples and actions taken. Avoid presenting automated scores as precise. The Academy's guide to a monthly social media report shows where sentiment fits.
A worked example
A restaurant tracks sentiment manually for three months. Overall, 70 percent of mentions with a clear tone are positive. Broken down by topic, food is 90 percent positive, while booking is only 40 percent positive, with repeated complaints about the online booking system. The restaurant fixes the booking flow and posts about the change. Over the next two months, booking sentiment rises to 75 percent. The topic breakdown turned a vague score into a clear action.
Sentiment and reviews
Review sites are rich sources of sentiment, often with more detail than social comments. Include them in tracking, especially for local businesses. Reply to reviews professionally, thank positive reviewers and address complaints calmly. UK rules prohibit fake reviews and misleading review practices; the Academy's guide to UK consumer rules explains what applies.
Sentiment for different industries
Some industries naturally attract more negative sentiment, such as utilities, transport and telecoms, where people mostly post when something goes wrong. Others, such as food, travel and fitness, attract more positive sharing. Compare your sentiment with your own past results and, where possible, with direct competitors rather than with brands in very different sectors.
Building a simple sentiment log
Create a spreadsheet with columns for date, platform, link, tone, topic and notes. Add each meaningful mention as you find it during weekly checks. Use a short, fixed list of topics, such as product, service, price, delivery and staff, so you can sort and count easily. After three months, the log becomes a valuable record of how customers feel and why.
Common mistakes
- Trusting automated scores without checking.
- Reporting a single score with no topic breakdown.
- Ignoring neutral mentions that contain questions.
- Treating negative sentiment only as a PR issue.
- Comparing sentiment across tools with different methods.
Learning sentiment analysis
The Academy's social listening module in the third level covers sentiment, share of voice and reporting, with practical exercises in labelling and interpreting mentions. The free assessment shows where you would start.
