Answers
How do you find the root cause of repeat contacts?
Stop looking at disposition codes and read what customers actually said. Cluster contacts by underlying reason, name each recurring cause, price it, and assign it to the function that owns the fix. Disposition data is too coarse to find root causes because it was designed for reporting, not diagnosis.
Why disposition codes don’t work
Agent-selected dispositions are chosen under queue pressure from a list someone designed years ago for a different purpose. They tell you a contact was about “Billing”. They cannot tell you that 380 of those contacts were one broken proration rule.
What works instead
- Read the reason, not the code. The customer’s own words — first message, subject line, transcript opening — are where the cause actually lives.
- Cluster by cause, not keyword. “Parcel says delivered but isn’t here” and “tracking shows delivered, nothing arrived” are one defect. Keyword grouping splits them; cause clustering doesn’t.
- Name each cluster. A named defect can be owned. A percentage in a category cannot.
- Price it. Volume × handle time × loaded cost. Ranking by cost settles the argument about what to fix first.
- Assign an owner outside the contact centre. Most root causes belong to billing, product, logistics or the web team.
- Re-measure after the fix. The step everyone skips, and the one that makes the second conversation easy.
The Failure Demand Taxonomy gives you a ready-made set of named defects to cluster against, so you are not inventing categories while you work.