How to Use AI to Analyse Customer Feedback

Customer feedback is one of the most valuable things a small business has, and one of the most ignored. It arrives in bits: a Google review here, a survey response there, a comment on Instagram, a grumble at the till. Each one is read, maybe replied to, and then forgotten. The patterns that could tell you what to fix or what to shout about stay hidden.

AI is excellent at finding patterns in lots of text. Paste in a pile of reviews or survey answers and it can group them into themes, count how often each comes up, pull out representative quotes, flag urgent issues and suggest what to improve first. You’ll still want to sanity-check its conclusions, but what used to take a weekend now takes an hour. This guide takes about an hour.

What you’ll need

• Customer feedback from the last few months: reviews, surveys, emails, comments
• A spreadsheet such as Google Sheets or Excel
• An AI chatbot and about an hour
• An open mind; some feedback stings

Step-by-step: analyse customer feedback with AI

1. Collect it in one place. Copy feedback into a spreadsheet with columns for date, source, star rating if any, and the comment. Even 30 to 50 comments is enough to see patterns.
2. Remove personal details. Delete names, emails, order numbers and anything that identifies a customer before using AI.
3. Find the themes. Paste the comments and ask the AI to group them into themes with counts and example quotes (prompt 1).
4. Separate praise and problems. Ask what customers love (use it in marketing) and what frustrates them (fix it).
5. Check the analysis. Read a sample of comments in each theme yourself. AI can lump things together or miss nuance.
6. Prioritise improvements. Ask the AI to rank problems by how often they come up, how serious they are and how easy they’d be to fix (prompt 2).
7. Act and close the loop. Make changes, tell customers what you’ve improved (prompt 3) and repeat the analysis every quarter to see if things shift (prompt 4). Google Sheets’ Explore and pivot tables help track ratings over time.

Copy-paste prompts

1. Find the themes

I run a [TYPE OF BUSINESS]. Here are [NUMBER] pieces of customer feedback with date, source, rating and comment (personal details removed): [PASTE].
Group the comments into themes. For each theme, give a short name, how many comments mention it, whether it's mostly positive or negative, and 2 representative quotes. Then list anything urgent, such as safety issues or repeated complaints about the same thing.

2. What should we fix first?

Here are the negative themes from our customer feedback with counts: [PASTE THEMES].
Rank them by how often they come up, how much they probably affect customers and how easy they'd be for a small business to fix. Suggest one practical action for each of the top three, and two quick wins we could do this week.

3. Tell customers what we changed

We've made these changes based on customer feedback: [CHANGES].
Write a short, friendly social media post and a two-line note for our email newsletter thanking customers for their feedback and explaining what we've improved. Keep it genuine, not salesy.

4. Compare this quarter to last

Last quarter's themes and counts: [PASTE]. This quarter's: [PASTE].
Compare them in a table. What's improved, what's got worse and what's new? Suggest what to focus on next quarter.

Recommended tools

• Claude: handles long lists of comments well and is good at nuanced themes.
• ChatGPT: can analyse uploaded spreadsheets on many plans.
• Gemini: works inside Google Sheets and Docs.
• Microsoft Copilot: helpful if your data is in Excel.
• Google Sheets: pivot tables and Explore for quick charts of ratings over time.

Tips and mistakes to avoid

• Protect customer privacy. Remove personal data before pasting it into any AI tool.
• Don’t over-trust the counts. AI counts can be slightly off. Use them as a guide.
• Look for the quiet issues. One or two comments about safety or accessibility matter more than their numbers suggest.
• Use the praise. Positive themes are ready-made marketing messages; quote customers with permission.
• Don’t take it personally. Negative feedback is a gift, even when it doesn’t feel like one.
• Ask better survey questions. Ask the AI to help write short surveys with one or two open questions.
• Track over time. A quarterly check shows whether changes are working.

FAQ

How much feedback do I need?

Even 20 to 30 comments can show patterns. More is better, but don’t wait for hundreds.

Can AI read my Google reviews automatically?

Some tools connect to review platforms, but copying them into a spreadsheet works well for most small businesses.

Should I reply to every review?

Replying to reviews, especially negative ones, shows you care. See our guide to replying to reviews with AI.

What if most feedback is positive?

Brilliant, and still useful. Look for what people praise most so you can protect it as you grow, and read the three- and four-star comments closely. They often contain the small suggestions that turn a good business into a great one.

I run a [TYPE OF BUSINESS]. Here are [NUMBER] pieces of customer feedback with date, source, rating and comment (personal details removed): [PASTE].
Group the comments into themes. For each theme, give a short name, how many comments mention it, whether it's mostly positive or negative, and 2 representative quotes. Then list anything urgent, such as safety issues or repeated complaints about the same thing.

Related guides

• How to use AI to reply to online reviews (good and bad)
• How small businesses can use AI to answer customer emails
• How to use AI to create a simple marketing plan

Reference: Google Docs Editors Help: Analyse data