A record created by a switch for each call attempt, capturing details such as calling and called numbers, start time, duration, route, and outcome.
CDRs are the system of record for voice networks. A typical record includes the calling and called numbers, the customer or trunk that originated the call, the route and carrier used, timestamps for setup, answer, and release, the call duration, and the release cause or SIP response code. Records are generated for failed attempts as well as completed calls, and those unanswered attempts are often where the most useful fraud signals appear.
They are used for billing, settlement with other carriers, capacity planning, and quality monitoring. Many providers also extend their CDRs to include STIR/SHAKEN details such as attestation level, origid, and verification status. Adding these fields at the source, rather than joining them in later from separate systems, makes analysis simpler and more reliable.
For fraud detection and compliance, CDRs are the raw material for spotting patterns: spikes to unusual destinations, short call bursts, numbers cycling rapidly, or traffic that does not match a customer's profile. Retaining them in a searchable form also makes traceback responses fast and accurate. Retention periods should cover at least the window in which traceback requests and disputes typically arrive.
Why it matters to carriers
When a traceback request arrives, you need to find the specific call, the customer, and the upstream source quickly. When fraud starts, you need to see it within minutes, not at the end of the billing cycle. Both depend on complete, timely, and searchable CDRs, so treat CDR quality as a compliance asset, not just a billing input.