Disposition codes are how a contact centre turns thousands of daily calls into structured, queryable data. The right code set reveals your pipeline health. A poorly designed one generates noise that managers ignore.
A call disposition code is a label — typically selected from a dropdown by the agent at the end of a call — that categorises the outcome of the interaction. Common codes include Interested, Not Interested, Callback Requested, Wrong Number, and Sale Closed. Dispositions feed CRM records, reporting dashboards, and dialler logic.
Most contact centres treat dispositions as a checkbox — agents click something so the call can be closed in the CRM. The problem with this approach: if agents select codes quickly or inaccurately, the downstream data is useless. Reporting on "Interested" rates or "Callback" follow-up rates only works if agents are selecting codes accurately and consistently.
In audited contact centres, studies consistently show that 15 to 30% of disposition codes are inaccurate — agents selecting "Not Interested" for calls that were actually "No Answer," or "Callback" for calls where the customer declined to speak again. AI transcript analysis can audit dispositions post-call to identify systematic mis-coding patterns.
| Code | Type | Meaning | Dialler Action |
|---|---|---|---|
| SALE | Sale | Customer committed, payment initiated or confirmed | Remove from active queue, move to onboarding |
| HOT | Callback | Interested, needs 24-48h to decide. Agent promised follow-up | Priority callback in 24h, same agent |
| CB_SCHED | Callback | Customer requested callback at specific time | Schedule callback at agreed time, same agent |
| CB_GENERAL | Callback | Vague callback — "call me later, busy now" | Generic callback within 48h, any agent |
| NI_FINAL | No Sale | Hard no — customer explicitly declined and asked not to be called | DNC flag, suppress permanently |
| NI_THINK | No Sale | Soft no — not interested today, may reconsider | Suppress 30 days, then re-queue |
| VM_LEFT | Neutral | Voicemail — message left | Callback in 4h max, do not leave second voicemail same day |
| NO_ANSWER | Neutral | No answer, no voicemail left | Retry in 2h, max 3 attempts per day |
| WRONG_NUM | No Sale | Wrong number — not the intended contact | Flag for data team to verify lead record |
| DNC | No Sale | Customer explicitly placed on Do Not Call | Permanent suppress, compliance flag |
| LANG_BARRIER | Neutral | Could not communicate — language mismatch | Route to regional agent queue |
| BUSY | Neutral | Line busy | Retry in 30 minutes |
| DROPPED | Neutral | Call dropped before outcome | Attempt callback within 10 minutes |
| ESCALATION | Callback | Complaint or issue — transferred to supervisor | Flag for QA review, manager follow-up |
Every call outcome should map to exactly one code. If agents hesitate between "Interested" and "Callback Requested" on the same type of call, your categories overlap. Define the distinguishing criterion explicitly in agent training documentation.
"Other" as a disposition is where data quality goes to die. If agents are regularly selecting "Other," it means you have real outcomes in your call mix that your code list does not cover. Audit your "Other" codes monthly and convert patterns into named codes.
A dropdown with 35 options takes 15 to 20 seconds for agents to scan. That is dead time multiplied across every call in the day. Each additional code also dilutes data quality — agents guess. Keep the code list to codes you actually act on differently.
A disposition code that does not trigger a specific follow-up action is decorative. Every code in your system should have a defined next action: suppress, retry at X, callback in Y hours, escalate to Z. If you cannot define the action, you do not need the code.
When AI transcribes every call, the transcript can be compared against the agent-entered disposition code. A call where the customer says "haan, mujhe interest hai, kal call karo" (yes, I'm interested, call me tomorrow) should be coded CB_SCHED — not NI_THINK or CB_GENERAL.
AI can flag calls where the transcript content does not match the disposition code, enabling QA teams to identify agents who are systematically miscoding — either to reduce their call count, pad "interested" rates, or avoid follow-up obligations.
Bolo Aur Likho transcripts create the text layer needed to audit dispositions: upload the call, compare the transcript against the CRM entry, and identify patterns across your team.
Upload any call to get an instant transcript. Compare what was said against what was coded. No signup required.
Try Free →