What is a disposition code?
A disposition code is the classification applied to a call once it has ended. It names what happened, such as a payment taken or a wrong number reached. Reporting and the decision to dial a number again are both derived from the code rather than from the conversation itself.
Also called: Call disposition, Outcome code
Why the taxonomy deserves more scrutiny than the agent applying it
Nobody reviews a full week of recordings, so the disposition code becomes the call as far as the organisation is concerned. Every retry rule and every figure presented to a board is derived from that single field, which means a code applied loosely produces a report that is confidently wrong. Automation changes the failure mode here, because an agent applies the taxonomy consistently and will therefore apply a flawed taxonomy to every call without variation. The taxonomy becomes the object needing review, rather more than the agent applying it.
- Downstream reporting counts codes rather than recordings, so the code becomes the call.
- A taxonomy with overlapping categories produces numbers nobody can reconcile.
- Consistent coding by an agent exposes how ambiguous the categories always were.
- Retry logic driven by codes will redial a wrong number repeatedly if coding is loose.
How a disposition code works
- The taxonomy is defined with each code mutually exclusive of the others.
- Each code names the evidence in the call that justifies applying it.
- The agent proposes a code from the transcript and the state reached.
- A confidence value accompanies the code, and doubtful cases are reviewed by a person.
- Codes are written to the system of record with the call reference attached.
- A sample is audited against recordings each period and the taxonomy amended.
How a disposition code is evaluated
A disposition taxonomy is judged on whether two reviewers would apply the same code to the same call. Measure the following on sampled calls:
- agreement between the agent code and a reviewer's code
- agreement between two independent reviewers on the same call
- share of calls landing in a catch-all category
- distribution across codes, with unused codes identified
- proportion of codes carrying the evidence that justifies them
Limitations, risks and common failure modes
- Catch-all categories absorb everything the taxonomy failed to anticipate.
- Codes inherited from a dialler describe outcomes the agent cannot observe.
- A code applied on intent rather than on outcome inflates success in reporting.
- Taxonomy changes are deployed unversioned, so historical reports stop comparing.