A good disposition set is short, mutually exclusive and tied to a next step. Each label (completed, no answer, callback requested, not interested, do not call) should trigger one clear action: retry, follow up, update the CRM or stop. In Edesy, the system sets line outcomes and the agent sets conversation outcomes, and campaigns retry only the dispositions you choose.
Set up outcome-based retries: Try Edesy free — ₹50 free credit, no card needed.
Why dispositions matter more with AI calling
A human team makes a few hundred calls a day and sorts them by memory. An AI campaign can make thousands, with 1 to 100 calls in parallel. Without clean dispositions you end up with a pile of "completed" calls and no idea which ones need a follow-up, which ones to retry and which ones must never be called again.
Dispositions are the routing layer of your campaign. Get them right and retries, CRM updates and reports follow on their own.
Two kinds of outcome
| Kind | Set by | Examples |
|---|---|---|
| Line outcome | The system, from what happened on the call | No answer, busy, failed, voicemail, transferred |
| Conversation outcome | The agent, from what the caller said | Completed, qualified, callback requested, not interested, wrong number, do not call, partial, no decision |
Line outcomes need no prompt work. Conversation outcomes do: the agent has to recognise the situation and set the outcome with the set call outcome tool. That is a mid-call action you attach to the agent and describe in the prompt.
The built-in set
Edesy agents can use these dispositions: success, voicemail, busy, no answer, callback scheduled, not interested, transferred, wrong number, do not call, partial, failed, qualified, completed, callback requested, no decision, plus custom ones. You will not need all of them in one campaign. Pick the ones your next steps depend on.
Four rules for a good set
1. Every disposition has one next step
Write the next step beside each label. If you can't, the label is not useful.
| Disposition | Next step |
|---|---|
| No answer / busy | Retry after a delay |
| Voicemail | Retry later, or leave a message (voicemail action) |
| Callback requested | Schedule a callback at the time the caller gave |
| Qualified | Update CRM stage, notify sales |
| Not interested | Update CRM, stop calling for this campaign |
| Wrong number | Mark the contact invalid; never retry |
| Do not call | Add to your suppression list; never retry |
| Transferred | Nothing automatic; the human owns it |
| Partial | Review; often retry once |
2. Labels don't overlap
"Not interested" and "no decision" are different: one closes the lead, the other keeps it open. Tell the agent exactly when to use each. A line in the prompt like "Use no decision only if the caller says they need to think or ask someone else" saves a lot of messy data.
3. Prefer custom outcomes for your business result
The built-in completed says the call finished normally. It doesn't say whether you got what you wanted. Add one or two custom outcomes for the result that matters: payment promised, site visit booked, order confirmed. These are what your weekly report should count.
4. Keep the stop labels strict
Do not call and wrong number must never be retried. Configure campaign retries to exclude them, and make sure your prompt tells the agent to set do not call whenever the caller asks not to be called again, in any language.
How dispositions drive campaigns
In the campaign wizard, retries are set per campaign:
- Attempts: 1 to 5.
- Delay: 5 to 120 minutes between attempts.
- Which dispositions to retry: typically no answer, busy and voicemail.
After a campaign, you can also retry from outcomes: pick a disposition such as callback requested or partial and start a new round for only those contacts. The daily calling window keeps retries inside the hours you allow. More detail is on the call retry and disposition page and in the campaign scheduling docs.
How dispositions drive workflows
Workflows can start on call ended and filter by outcome. Typical rules:
- Qualified → CRM update, WhatsApp to the lead, email to the sales owner.
- Callback requested → schedule callback, or book a calendar slot.
- Not interested → CRM update with the reason from extracted data.
- Do not call → webhook to your suppression list.
The call.ended and call.analyzed webhook events carry the outcome if you route it from your own
system instead. Pair the disposition with extracted fields, such as the callback time or the reason for
no interest; see voice call data extraction to CRM.
Common mistakes
- Too many labels. Twenty dispositions means the agent picks inconsistently. Merge anything that shares a next step.
- Retrying everything. Retrying not interested annoys people and hurts answer rates on future campaigns.
- No custom success label. You can't tell a good campaign from a bad one if every finished call is completed.
- Forgetting partial calls. Calls that dropped mid-way need their own label and usually one retry.
- Not testing outcomes. Add eval cases that check the agent sets the right outcome; see testing voice agents with evals.
A starter set for outbound reminders
For an EMI or appointment reminder, this set covers most calls: no answer, busy, voicemail, completed, payment promised (custom), callback requested, not interested, wrong number, do not call, transferred, partial. Retry no answer, busy, voicemail and partial; follow up callback requested; stop on the rest. The EMI collection and appointment reminder templates come close to this out of the box.
What it costs
Dispositions, retries and outcome-based workflows are included. Calls cost from ₹4/min on plans and credit packs (₹6/min pay as you go), plus telephony; see the pricing page. New workspaces get ₹50 free credit on sign-up, no card needed.