AI System for Grouping Customer Feedback into Product Themes

The problem

In day-to-day operations, surveys, support records and permitted interview notes arrive through different systems and files, preventing the team from reviewing the same case in one view.

As a result, the link to source evidence can be lost while source-linked themes and examples are prepared, leading to repeated checks and uncertainty about the basis for the decision.

Incomplete automation may overlook an isolated request presented as a broader trend, while incorrect automation risks obscuring minority needs.

What we set out to improve

• Time taken for the draft of source-linked themes and examples to be ready for review
• Correct routing of records where an isolated request appears to represent a broader trend to the appropriate specialist queue
• Record of suggestions corrected or rejected by the product researcher

How the system works

1. 01 Retrieve surveys, support records and permitted interview notes from authorised sources
2. 02 Verify the source system, record identifier and recency information
3. 03 Map the fields to the target process schema
4. 04 Prepare the draft of source-linked themes and examples
5. 05 Compare the draft with business rules and source records
6. 06 Check for an exception: an isolated request appears to represent a broader trend
7. 07 Have the product researcher approve, revise or reject the suggestion
8. 08 Write the input, suggestion, changes and final decision to the audit log

Methods we used

• Source-traceable data extraction for surveys, support records and permitted interview notes
• Structured output that limits the source-linked theme and example fields through business rules
• An exception gate that separates cases where an isolated request appears to represent a broader trend
• A queue that shows the product researcher the source evidence and suggestion together
• Duplicate-processing controls and a reversal record to reduce the risk of obscuring minority needs

Where people stay involved

Data and security

Only fields required for the task are processed from surveys, support records and permitted interview notes. Source-system permissions are preserved, and personal or commercially sensitive data is minimised before being passed to the model. Every read, suggestion, human change and write to the target system is logged.

Who this suits

Frequently asked questions