Predictive Dialer Abandonment on a Loan Campaign
You supervise a 40-seat outbound campaign for a bank's pre-approved personal loan in Cuttack. The predictive dialer runs at a 3:1 ratio between 10:00 and 19:00. Last week's reports show an abandonment rate of 6.2 percent against a regulatory cap of 3 percent per campaign per day, 11 percent of connects dropped within two seconds, average wait after the customer says hello is 4.6 seconds, and 2,300 numbers were dialled twice in the same day. One agent reports that before her call connected she faintly heard the previous customer's conversation in her headset. Two customers on today's list state they are registered under Do Not Disturb. The bank's contract carries a penalty clause for compliance breaches. Today's calling list has 9,400 numbers, of which 1,100 were already contacted in the last 48 hours.
1. Diagnose why abandonment is above the cap and give three concrete changes, in priority order.
Read the worked answer
Diagnosis first. Abandonment rises when the dialer predicts more answered calls than there are free agents. The 4.6-second average wait after hello and the 11 percent of connects dropped within two seconds point to over-dialling, not to customers hanging up. Contributing causes: agent logins not matching the roster through the day, long wrap/after-call work holding agents out of the ready pool, and a fixed 3:1 ratio that stays high during lunch and low-staff windows. Repeat dialling of 2,300 numbers in one day also inflates connects per hour without adding agents.
Priority changes:
- Move to progressive dialling (or drop to 2:1) during the 13:00–14:00 and 17:00–18:00 troughs; predictive ratios are safe only when staffing is predictable.
- Re-sync the dialer to logged-in ready agents every 15 minutes instead of once per shift, so the algorithm stops chasing agents on break or in wrap.
- Cap wrap-up at 30–45 seconds and suppress repeat dials of the same number within 24 hours; the 2,300 double-dials are pure waste.
Reasoning: the 3 percent cap is measured per campaign per day, so a single bad hour breaches it for the whole day. Control the troughs and the ratio can stay aggressive at peak.
2. What do you do about the agent who heard another customer's conversation, and about the two Do Not Disturb claims?
Read the worked answer
The overheard audio is a potential privacy incident and is treated as such, not as a headset glitch to shrug off.
- Ask the agent to stop taking calls, and take a written statement of exactly what she heard, when, and on which extension - without repeating the customer content to the floor.
- Log a ticket with IT/telecom: possible causes are channel bleed on an analogue headset, cross-talk on a shared trunk, or a recording/whisper configuration error. Preserve the dialer logs and recordings for that window before they roll over.
- Check whether other agents on the same bank of extensions heard anything; if this is systemic it is a data-protection breach that must go to the client's compliance contact, not just the internal IT queue.
- Do not identify or contact the affected customer unless the client's compliance team directs it; the agent must not discuss it on the floor.
The two Do Not Disturb claims are handled separately and immediately: apologise, do not argue or try to retain, mark opt-out in the CRM with a timestamp, and scrub them from the list before the next dial cycle. Then check why they were on the list at all - usually a stale list upload or a filter not applied to the dialer's import.
Reasoning: opt-out is a consent issue with statutory backing; an argument on the phone creates a complaint where a one-line apology would have closed it.
3. A customer says, 'Stop calling me, I never applied for any loan.' Script your response and the CRM action.
Read the worked answer
Response, in this order:
- Apologise once, plainly: 'I am sorry for the disturbance, sir. I will make sure this number is not called again for this campaign.'
- Do not defend the data source on the call, do not ask him to stay on the line for 'just two minutes', and do not transfer to a retention desk. Any of these converts an opt-out into a complaint.
- Take the opt-out at the number level, not just at this record: confirm the number he is speaking from so the wrong entry is not marked.
- Close politely and let him hang up first.
CRM action: set the contact preference to Do Not Call - customer request, with the date, time, campaign code and agent ID; the flag must apply across campaigns, not only this loan product. Raise a data-source query if he genuinely never applied, because that suggests the list was bought or appended, which is a client-side compliance issue.
Reasoning: the words 'never applied' are a data-source allegation as well as an opt-out. Recording both protects the customer and gives the client the evidence it needs to audit the vendor who supplied the list. An agent who argues loses the customer and the audit in the same call.