Analyzing customer complaints from contact recordings: causes table

Call counts and issue counts differ

One customer calls three times about one order, while another sends one message. Counting messages without grouping them by case gives the first issue more weight just because of follow-up contacts. Those contacts are still useful evidence: they may show that the answer or resolution did not meet the need.

Before analyzing, choose the period, channels and goal. For example, find out why customers keep asking when a service will be ready. Keep pre-purchase inquiries, questions during use and complaints about the outcome distinct, since each calls for a different action.

Build a table that helps the team choose an action

1. Group comparable cases

   Define the inclusion rule and group related contacts using an available work identifier. Keep the count of follow-ups separately. Avoid copying unnecessary contact details. An authorized internal link or code is enough to find the source again.

2. Tag what you can observe

   Record the stage where the difficulty arose, what the customer expected, what happened and how it ended. Check important facts in the work system. Treat a remark about an event as a customer account until available data confirms the event.

3. Group and verify causes

   Group cases by the mechanism behind the problem. “I didn't understand the status” and “the task was actually overdue” can sound similar but call for different fixes. For each suspected cause, record supporting evidence, counterexamples and missing information.

4. Choose an improvement and a measure

   Specify a change, an owner and how to check it. If you compare contacts before and after, use the same counting rules and account for changes in service volume. A drop in complaints alone does not explain the cause without context.

Repeated questions about when the work will be ready

This is a fictional training example. The names, circumstances and outcomes are included solely to explain the method.

The symptom is a follow-up request for status. The message may fail to explain whose action is needed. First check whether the work is on schedule and whether the unclear status is masking a real delay.

For this example, test a notification that states the current stage and expected action. Compare equivalent cases to measure the result without promising a made-up percentage drop in contacts.

Customer contact analysis table

Start with a small, comparable sample. Refine the categories, then tag the entire selected period.

Period / channels / inclusion rule:
Unit of analysis:
Case code / number of contacts:
Customer journey stage:
Customer expectation:
Symptom and consequences:
What the system confirms:
Timecode or source:
Suspected cause:
Alternative explanation:
Change / owner:
How and when to check the outcome:

Check that the comparison is fair

• Follow-up contacts are linked to the original case.
• Emotional wording is kept separate from what happened.
• The cause has supporting evidence or is labeled as a hypothesis.
• Periods are compared using the same rules.