Classify customer feedback sentiment by theme

Step 1: Isolate the feedback columnExcel

Convert the data to an Excel Table with the open-text feedback in one column and a unique row ID in another (a simple sequential number works) — the ID is what lets you trace any specific classification back to the original comment later, and the Table keeps both columns aligned as rows get sorted or filtered.

Step 2: Ask Copilot to classify and scoreCopilotExcel

Give it a fixed theme list up front rather than letting it invent categories on the fly — otherwise you'll get 40 near-duplicate themes instead of a usable pivot.

Prompt idea:

For each row in the Feedback column, add a Theme column classifying it into exactly one of: Quality, Pricing, Service, Packaging, Other. Add a Sentiment column scored -1 to 1. Don't invent new theme categories.

Step 3: Pivot the resultsExcel

Build a pivot table of theme by sentiment to see which themes are both frequent and negative — that's where to focus.

Step 4: Spot-check the "Other" bucketExcel

If Other is more than ~10% of rows, your theme list is missing something real — read a sample and consider adding a category.

Check the work

• Read 15-20 classified rows against their assigned theme and sentiment to confirm they're reasonable.
• Check that ambiguous feedback (mixed sentiment) wasn't force-fit into a single bucket.
• Confirm the "Other" category isn't hiding a real, common theme.

Source: Nexacu, "Copilot for Excel (2026): Practical Prompts, Agent Mode & Time-Saving Examples" (2026)