Visibility from INVITE to BYE.
Use call records and quality indicators to understand traffic behavior and investigate routing issues.
| Destination | Attempts | ASR | ACD | Cost | Status |
|---|---|---|---|---|---|
| Demo group A | 8,240 | 62.1% | 04:42 | $48.20 | Normal |
| Demo group B | 5,180 | 61.3% | 04:28 | $36.10 | Review |
Use call records to explain the totals.
Connect a metric to its definition
A CDR is useful only when its fields and timestamps are understood. Identify attempt time, answer time, disconnect time, duration and the route or trunk references available in your reporting. A billed duration may differ from the raw elapsed duration because of the agreed charging increment.
Compare time zones and reporting windows before reconciling totals. Separate call attempts from answered calls, and distinguish end-user calls from carrier attempts when retries are involved.
Read ASR, ACD and PDD together
ASR and ACD reflect both network delivery and the behavior of the traffic being sent. PDD helps investigate setup delay, but its interpretation depends on where measurement begins and ends. NER requires an agreed definition of which outcomes count as network-effective.
CPS and concurrent calls describe different aspects of load. Use them alongside completed minutes and call outcomes to explain a busy period. Confirm export fields, retention, access and any automated retrieval options with the account team before designing downstream reporting.
Where it fits in your network
- 01Call attempts
- 02CDR generation
- 03Quality analysis
- 04Operational review
Plan the connection around your traffic
Read metrics together
ASR measures answered attempts, ACD measures answered-call duration and PDD describes setup delay. Traffic and destination mix affect all three.
Investigate by segment
Start with time window, destination and trunk. Compare SIP responses and individual records before drawing conclusions.
Reconcile usage
Agree time zone, duration units, billing increments and export format. Match sample records to an invoice before automating reconciliation.
Agree the details before production
| Parameter | Planning detail |
|---|---|
| Exports | Confirm formats and access methods. |
| Retention | Confirm contracted retention period. |
| Quality | Review ASR, ACD, PDD and response definitions. |
| Privacy | Restrict CDR access to authorized staff. |
What to review with your team
| Review area | What to establish |
|---|---|
| Definitions | Attempt, answer, duration, chargeable time and NER treatment. |
| Comparison | Consistent destinations, time zones and traffic groups. |
| Reporting | Field availability, retention, permissions and retrieval scope. |
Before the next step
Why do my totals differ from a carrier report?
Check reporting windows, time zones, answer criteria, retries and billing increments. Compare a small set of matching records before investigating the full total.
Can I retrieve CDRs through an API?
Confirm the supported account-specific specification and access permissions. This site does not publish an unverified API operation.
Start with the question the report needs to answer
A quality report, an incident report and an invoice reconciliation use some of the same call records but answer different questions. Quality analysis compares outcomes for a traffic group. Incident analysis follows a particular failure. Reconciliation explains how recorded usage became a charge. Decide which question is being asked before choosing an aggregate.
Document the meaning of every field used in a calculation. Answer time, release time and billable duration may not be interchangeable. A reporting system that silently mixes time zones or counts every retry as a separate customer call can create a misleading trend.
Use dimensions before conclusions
Break an aggregate by destination, trunk and time window. Add other dimensions only when they help isolate a difference. Small samples and changing call patterns need context; a percentage without its attempt count can be difficult to interpret.
For an unexplained change, select a few representative records and follow the signaling outcome. Compare a successful call from the same period where possible. The troubleshooting workflow helps turn a dashboard observation into evidence that another team can investigate.
Design the reporting handoff
Confirm export fields, retention and account permissions before building a downstream process. If an automated interface is available, use its supplied specification and error behavior. Keep personal calling information and credentials out of demonstration datasets.
For billing analysis, preserve the rate version and chargeable-time rules alongside the record. The billing example shows why a nominal rate alone does not explain the final amount. Reporting should make the commercial calculation traceable, not merely reproduce a total.
Turn a billing discrepancy into a small comparison
A monthly total differs from an internal report. Start by aligning the time window, timezone, currency and record inclusion rules. Then select a few answered calls that show the difference. Compare recorded duration, chargeable duration, rate and increment rules rather than attempting to diagnose the whole invoice at once.
A 61-second call may be rounded differently under different agreed increments. Retries and failed attempts can also be counted differently in operational reports. Explain those definitions before concluding that records are missing. The worked billing example in the documentation hub provides the arithmetic; actual charges follow the service’s commercial terms. This is a reconciliation example, not sample customer billing.