Why 85% of Collection Calls Go Unanswered
Spam filtering, screening, bad timing, and bad data mean 85% of outbound collection calls go unanswered. Pay-per-answered AI makes it cost nothing.
TL;DR
85% of outbound collection calls go unanswered. Debtors screen unknown numbers. Spam-blocking apps filter out collections prefixes. Call times drift off optimal windows. Human call centres pay agents to listen to dial tones for four of every eight hours. AI voice agents eliminate this line item: pay-per-answered-call economics mean unanswered attempts cost near-zero. But the deeper insight is that the 85% unanswered rate itself can be reduced through smarter caller ID handling, time-of-day optimisation, and channel blending. This post breaks down the economics and the tactical levers.
Why So Many Calls Go Unanswered
- Spam filtering. Apps like Hiya, Truecaller, and carrier-level spam filters categorise high-volume outbound numbers as likely spam.
- Unknown-number avoidance. 60-70% of consumers report ignoring unknown numbers by default.
- Bad timing. Traditional diallers call during working hours when the debtor is at work and unable to answer.
- Disconnected numbers. Portfolio data ages; 5-15% of numbers are no longer valid.
- Deliberate avoidance. Debtors actively screen creditors they are avoiding.
The Economic Impact on Human Operations
A UK collections agent on shift makes 100-120 dial attempts per day. Of those, ~15-20 connect. The other 80-100 attempts produce dial tones, voicemails, and spam-blocked disconnects. The agent is paid for all of them.
Annualised at a 50-agent operation with fully-loaded cost of GBP 42,000 per seat, that unanswered volume represents GBP 1.5-2 million per year of pure wasted salary. No human pick-up happens on those calls. No outcome is produced. Yet the cost shows up in the P&L.
Stat block: unanswered economics
- 85%: Typical outbound collection non-answer rate.
- 4-5 hours: Per shift that human agents spend listening to dial tones.
- GBP 1.5-2m/year: Wasted unanswered-call labour cost at a 50-agent operation.
- ~0p: AI voice agent cost per unanswered attempt under pay-per-answered model.
Why Pay-Per-Answered-Call Changes the Math
With an AI voice agent on pay-per-answered-call economics, the 85% unanswered rate costs virtually nothing. You pay for the 15% that connect. The AI handles those with top-performer consistency and structured evidence. The 85% that do not connect are retried, rescheduled, or suppressed based on pattern. See pay-per-answered ROI framework.
How to Actually Reduce the 85%
Beyond the economic reframe, AI operations can meaningfully reduce the non-answer rate:
- Caller ID reputation. Use verified, low-reputation-risk numbers rather than predictable long strings.
- Time-of-day optimisation. Learn per-debtor optimal windows. Some debtors answer lunchtime, others evening.
- Channel blending. Pre-warm with SMS. Debtors who see a friendly SMS answer the subsequent call.
- Cadence discipline. Three attempts in one day trains debtors to block. Spacing raises answer rates.
- Portfolio hygiene. Suppress disconnected numbers quickly to avoid dragging down answer rate metrics.
What Debtors Actually Want
Post-answer survey data shows debtors appreciate:
- Clear identification of who is calling and why (not an anonymous 030 number).
- Respect for stated callback times.
- A channel choice (some want SMS not call).
- Short, focused conversations when they do answer.
AI voice agents can honour all four consistently. Human operations try but drift under volume pressure.
Bottom Line
The 85% unanswered rate is not a failure. It is a structural feature of outbound collections. The question is whether you pay for it in wasted agent hours or whether you restructure the economics so it costs nothing. AI voice agents do the latter. See related: cost comparison, ROI framework, propensity prediction.
Call Sarah on +1 (332) 241-0221 or book a consultation.
Frequently Asked Questions
Does AI have the same spam-filter issue as human diallers?
Yes, inbound carrier spam filtering applies regardless of who is calling. Caller ID reputation, verified numbers, and branded caller display are the countermeasures used in AI deployments.
Can AI really learn per-debtor time-of-day patterns?
Yes. After 2-3 call attempts, patterns start to emerge. After 5-8, they stabilise enough to drive scheduling decisions.
What about the cost of telephony itself on unanswered calls?
Telephony is charged on connected minutes, not attempted calls, in most modern architectures. The economic impact of unanswered attempts is negligible.
Does the AI deliver voicemails?
Only where permitted and policy-configured. Some jurisdictions restrict AI-voice voicemail delivery.
How does this affect the creditor's perception of effort?
Creditors receive transparent reporting on total attempts, connect rate, and outcomes. Perceived effort tracks actual effort because the data is auditable.