How to analyse customer feedback: from calls and complaints to concrete decisions
Sado AI team · · 4 min read
In short
Analysing customer feedback means counting what customers say in calls, complaints, surveys and reviews and turning it into decisions. First set 6–10 categories: price, delivery, quality, staff behaviour, waiting for an answer and so on. Put each piece of feedback into one category and count weekly. Pick the three most frequent problems and give each an owner and a deadline. A month later, check whether they have dropped. Listening to ten calls yourself every week shows you the reasons behind the numbers.
Where customer feedback lives
- Call recordings and transcripts, the biggest source: customers say in their own words what they didn’t like.
- The complaint log: what happened and how it was solved. On the log: complaint handling.
- Surveys: NPS and CSAT scores and the answers to “why?”.
- Reviews on Instagram, Telegram and maps.
- What salespeople and couriers hear, if they write it down.
In many companies these sources sit apart and nobody sees them together. The analysis starts by gathering them into one sheet.
Customer feedback analysis in six steps
- Set the categories: 6–10, clear to everyone. For example: price, delivery, quality, staff behaviour, waiting for an answer, payment, product range, other.
- Put each piece of feedback into one category with a short note: “courier two hours late”, “price differs from the website”.
- Separate positive from negative. Praise matters too: once you know what people like, you won’t break it.
- Count weekly and compare with the previous week. A one-week jump is a signal; growth three weeks in a row is a problem.
- Pick the three most frequent negative categories and give each an owner and a deadline.
- Check a month later: has feedback in that category dropped? If not, the cause lies elsewhere.
Example: a one-month table
| Category | Negative | Sample comment | Decision |
|---|---|---|---|
| Delivery | 46 | “The courier didn’t come at the time given” | Couriers give a two-hour window and call ahead if late |
| Waiting for an answer | 31 | “Nobody picked up at lunchtime” | Someone on duty or an AI operator at lunch |
| Price | 18 | “The website shows another price” | Prices checked every Monday |
| Staff behaviour | 7 | “He was rude” | Listen to the recordings and talk to the employee |
These numbers are only an example. A table of three or four rows shows a manager in half an hour which decision is needed.
Check the numbers against recordings
A number shows what there is a lot of, but not why. Every week, listen to 3–5 calls from each main category. Sometimes one recording reveals the cause: for example, half the complaints about delays might come from one district.
If you can, tell the customers who complained what has changed. That way they know their feedback wasn’t wasted.
Common analysis mistakes
- Counting only complaints. People who call with a question give feedback too: “where can I find the price?” means the price is hard to find on the website.
- Changing the categories every month. Then you can’t compare one month with another.
- Deciding by the loudest customer. One angry call doesn’t outweigh a hundred quiet customers: count first.
- Analysis without decisions. The table is made, but nobody gets a task, so next month it all repeats.
Collecting feedback in the Sado portal
With Sado’s AI operator, every call in the portal has a recording, full transcript, short summary and the answers to your questions. For example, the operator can ask at the end of a call, “What could we do better?”, and the answer is saved as a separate field.
You download the results to Excel and count them by category. You can ask the portal assistant «Yordamchi» about calls in plain words, for example: “What did customers say about delivery this week?” Surveys run through auto-calls. Try it live or call 78 778 01 23.
Questions
How often should you analyse customer feedback?
A short count every week and a deeper review once a month are enough. After a big change, such as moving to a new delivery service, look every day.
What if there are too many categories?
Beyond ten, counting gets confusing. Merge rare ones into “other”, and if “other” grows, split a new category out of it.
Who should do this work?
One person in charge, for example the head of customer service. Decisions are made by each category’s owner: delivery by logistics, price by sales.