How to read sentiment in a selected set of customer reviews
Sentiment analysis helps you see how customers describe their experience across a review sample. It is useful for organizing feedback before a closer reading, especially when praise and frustration appear in the same comment.
Start with the right material.
- What you provide
- a bounded visible set of customer reviews.
- The intended result
- Label each review positive, neutral, mixed or negative with a reason and uncertainty.
How the workflow works.
Classify the wording of each review and preserve a short explanation. Keep mixed sentiment visible, and distinguish a complaint about delivery from a judgment about the product itself.
What to check in the result.
Inspect ambiguous language and compare the labels with the original comments. The selected reviews are a sample; their balance should not be presented as the opinion of every customer.
Import the CSV with a consistent delimiter and inspect text identifiers, dates and amounts before using spreadsheet totals. Keep the original file available while checking any automatic type conversion.
Learn from the worked example.
The downloadable CSV example uses these columns: Review, Sentiment, Reason, Confidence. Compare a complete record with the source before sorting, combining or importing your own output.
The supplied example is illustrative. It is not a recorded run on your data.
Download the example ↓Use it with your own material.
- Open the source you want to work with and activate WebAct on that tab.
- Copy the task from the hero, add any required fields or preferences, and review the proposed actions.
- Compare the first result with your source. Ask for a specific correction or refinement where needed.