How to measure repetitive customer questions

Collect a representative sample

Review a recent period of email, chat, form and call notes. Start with enough conversations to include normal weeks and common seasonal peaks. Remove personal details from the working sample and record the customer's underlying question rather than copying whole messages.

Group by intent

Different wording often expresses the same need. 'Where is my certificate?', 'When do certificates arrive?' and 'Can I download proof of completion?' may belong to one certificate-access topic with a few distinct answer conditions.

• Name each topic in plain customer language.
• Count occurrences and the channels where they appear.
• Mark whether a published, current answer already exists.
• Separate general guidance from questions that require an individual decision.

Prioritise by value and answerability

The best first topics are frequent, time-consuming enough to matter and answerable from stable guidance. A high-volume question that always requires account access or judgement may need a person; a slightly less frequent policy question with a clear published answer may be a better self-service candidate.

Measure outcomes honestly

Track grounded answers, fallbacks, explicit visitor feedback and contact-link referrals separately. Do not call a conversation resolved simply because the visitor stopped typing. Review whether the assistant used the intended source and whether the source itself needs improvement.