Product feature
Measure reply speed with conversation context
Review customer service response times by agent and workspace, and investigate slow or unanswered conversations with the original evidence.
Measure the right event
A customer message, an ownership event and a successful CS reply are different events. A useful report keeps those timestamps separate. A Handle event is responsibility evidence; it is not automatically the same as the first customer reply.
Read the report in context
iTGTracker provides a separate QA response-speed area. Use the workspace and reporting period to compare agent counts, average time and the available delay categories. Confirm the metric definition and thresholds in the report before comparing results across teams.
Investigate the slow cases
An average can hide a small number of very slow conversations. Review the underlying conversation evidence and the ownership timeline. A handover, missing capture, non-working period or unresolved customer identity may change how you interpret a delay.
Keep unanswered messages visible
The QA Unanswered area provides a separate place to inspect captured customer messages without a corresponding response. Review the evidence before deciding whether a message required a reply. Do not substitute a quality score for a response-time measurement.
Collection and reporting limits
Authorised tracked Telegram conversations are collected incrementally about every 15 minutes; ended service rounds may be collected earlier. This collection interval does not mean AI scoring runs every 15 minutes. Tracker is needed for live CS attribution, and reporting completeness depends on successful collection.