ASR drops in a report and the manager asks to switch routes. That might be the right move, but one number does not prove it. It combines calls at different times, to different networks and different people. Start with how it was calculated.

Read the definitions

ASR commonly means the share of attempts with an answer signal. ACD is the average duration of answered calls. Ask whether very short answers count and how forwarded calls are treated. Voicemail can return an answer signal without a conversation with a player. Telephony cannot know whether a call became a useful contact unless an agent or CRM records the outcome.

Split the graph into comparable groups

Examine markets, destination networks, caller IDs and dialing times separately. A drop limited to one country after a caller ID change calls for different checks than a drop across all destinations. Compare like campaigns and periods. If the contact list changed, do not attribute the result to routing without more evidence.

Open specific call records

Take several cases from the affected group. Match time, status, duration and CDR with the agent's observation. Ask for a technical review if an answered call may not have been a real conversation. Authorized test calls to the relevant networks can show whether the issue is reproducible now.

Proton analytics provides ASR and ACD by destination. Use detailed call records alongside charts to investigate individual calls. A metric is useful when you can open an individual call and decide which hypothesis to test next.

Want to test calls in your markets?

Tell us which countries, numbers and call flows matter to your team. We can discuss a pilot with real test calls.

Discuss a call test →