Running an Outbound Dialer: Pacing, Abandonment and the Levers That Matter | HL Hunt
Running an Outbound Dialer: Pacing, Abandonment and the Levers That Matter
A dialer exists to remove the time an agent spends waiting, and the price is calls that connect with nobody available. Twenty-two agents at a 2.4 pacing ratio produce 739 attempts an hour and about 166 abandoned calls a day, a rate of 2.8% sitting inside a 3% cap. Both framings are accurate and they produce different decisions. The more useful finding is that pacing is the weakest lever available, and four others raise contact rate without costing any abandonment at all.
What you'll learn
The dialer trade, stated as arithmetic
A dialer exists to remove the time an agent spends waiting. The cost of doing that is calls that connect with nobody available to take them.
Take 22 agents who can handle about 14 contacts an hour manually:
- Manual capacity: 308 contacts an hour
- Predictive dialing at 2.4 attempts per agent: 739 attempts an hour
- Abandonment at 2.8%: about 20.7 abandoned calls an hour, or 166 a day
Both framings are accurate and they produce different decisions. Per our quality monitoring analysis, a rate expressed against a cap invites managing to the cap, while the same rate expressed as a daily count of people invites asking whether the throughput is worth it.
Reducing the ratio to 1.8 cuts attempts to about 554 an hour and abandonment proportionally. That is a real loss of throughput, and whether it is worth paying depends on what the extra attempts are actually producing, which is the next question.
What the extra attempts are worth
Raising the pacing ratio does not raise contacts proportionally, because the constraint moves.
- Agents become the bottleneck. Beyond the point where agents are continuously occupied, additional attempts produce abandonment rather than conversations.
- Right-party contact rate governs everything. Per our contact analysis, dialing more numbers for the same account reaches the same wrong number more often.
- The marginal attempt is the worst attempt. Attempts are made in priority order, so each additional one is on a less promising record, per our prioritization analysis.
- Repeat attempts to the same number decay quickly. The second call to a number that did not answer is worth far less than the first, and the fifth is worth almost nothing.
Point three is the one that breaks the intuition that more attempts means more recovery. If the work list is ordered properly, the attempts you add by raising the ratio are the ones your own scoring placed last, so the throughput gain is concentrated in the records least likely to produce anything.
Point four argues for breadth over depth. Per our customer view analysis, attempt limits should be per customer rather than per number, because a customer with four recorded numbers can receive four times the intended contact while every individual number stays inside its limit.
The constraints that are not negotiable
| Constraint | Why it has to be a hard gate |
|---|---|
| Permitted calling hours, in the consumer's time zone | Area code is not location, and a mobile number tells you nothing about where the person is |
| Attempt and contact frequency limits | Per our customer view analysis, these are customer-level, not number-level |
| Cease contact and channel restrictions | Per our cease contact analysis, a suppression that reaches the list slower than the list runs is not a suppression |
| Abandonment caps and message requirements | Where an abandoned call must play a message, that is a requirement rather than a courtesy |
| Reassigned and wrong numbers | Calling a number that no longer belongs to the debtor is a separate problem from calling too often |
| Consent status by channel | Consent for one channel is not consent for another |
The first row is the operational error that catches well-run shops. Time zone derived from area code was approximately right when numbers were geographic and is now frequently wrong, so a list dialed on area code will place calls outside permitted hours for a predictable share of records, every single day, with nothing in the dialer reporting it.
The fifth row is the one that produces the most serious individual exposures. Per our accuracy analysis, a wrong number repeatedly dialed is a person with no relationship to the debt receiving collection calls, and the only defence is a process that removes the number on the first indication rather than after a complaint.
Pacing cannot be the only lever
An operation whose only dial strategy is the pacing ratio will manage abandonment by moving that number up and down and nothing else. The levers that matter more are which records are attempted, how many attempts per customer rather than per number, what time of day each record is attempted, and whether the number is still correct. All four improve contact rate without raising abandonment at all, and all four are ignored when the ratio is the only control anyone touches.
Time of day is the free improvement
Contact rate varies substantially by time of day and by day of week, and it varies differently for different populations.
- Attempt each record at the time its own history suggests, not at the time the list reached it.
- Record the time of every successful contact and use it, which per our institutional memory analysis is information most operations collect and discard.
- Vary the time across attempts. Four attempts at the same hour is one attempt repeated.
- Respect stated preferences, which are both a requirement and the highest-contact-rate time available.
- Per our irregular income analysis, shift workers are unreachable at the times most lists are worked, which means a population is systematically under-contacted by a schedule nobody chose for that reason.
The third bullet is the cheapest improvement in this article. Rotating the attempt time across a record's attempts raises contact rate with no additional attempts, no change to pacing, and no abandonment cost, and most dialers will do it on configuration.
Throughput that does not come out of abandonment
HL Hunt AI Debt Collection applies attempt limits at customer level rather than per number, derives permitted calling hours from verified location rather than area code, rotates attempt times using each record's own contact history, and applies suppression in the same cycle it is received rather than the next one.
What to report
| Metric | Why |
|---|---|
| Abandoned calls as a daily count, not only a rate | Because the rate manages to a cap and the count describes people |
| Right-party contacts per agent hour | The actual output; attempts and connects are inputs |
| Contacts per customer against the limit | Catches the per-number loophole |
| Calls placed outside permitted hours | Should be zero; is not zero where hours come from area code |
| Wrong-number identifications and time to removal | The gap between the two is the exposure |
| Contact rate by attempt time | The free throughput that pacing cannot provide |
| Suppression lag | Time from a cease request to the number leaving every active list |
The second row is the only output measure on the list. Per our prioritization analysis, attempts, connects, and talk time are all inputs, and an operation optimizing them can raise every one while right-party contacts per agent hour falls.
The last row is the number nobody has and everybody should. A cease contact request that takes until tomorrow's list build to take effect means calls placed after a request to stop, which is the single clearest conduct failure available in this area and is produced entirely by a batch schedule rather than by any decision.
Measure the output, not the inputs
Get started with HL Hunt AI Debt Collection for right-party contacts per agent hour as the headline figure, abandoned calls reported as a count alongside the rate, outside-hours calls tracked as a control breach, and suppression lag measured to the cycle.
Frequently asked questions
At a 2.4 ratio, 22 agents produce about 739 attempts an hour and 166 abandoned calls a day at a 2.8% rate, which is inside a 3% cap.
The marginal attempt is on the record your own scoring ranked last, and repeat attempts to an unanswered number decay fast.
Deriving calling hours from area code, which is no longer location. Also applying attempt limits per number instead of per customer.
Record selection, per-customer attempt limits, attempt timing, and number accuracy. Rotating attempt times is the cheapest and costs no abandonment.
Key takeaways
- A 2.4 pacing ratio on 22 agents gives about 739 attempts an hour and 166 abandoned calls a day at a compliant 2.8%.
- Report abandonment as a daily count as well as a rate, because a rate invites managing to the cap.
- The attempts added by raising the ratio are the records your own scoring placed last, so the throughput gain is the least promising work.
- Attempt limits must be per customer, not per number, or four recorded numbers deliver four times the intended contact.
- Permitted calling hours derived from area code place calls outside those hours every day, invisibly.
- A reassigned number repeatedly dialed reaches someone with no relationship to the debt, and the defence is removal on first indication.
- Rotating attempt times using each record’s own contact history raises contact rate at no abandonment cost.
- Suppression lag is the clearest conduct exposure here and is produced by a batch schedule rather than any decision.
This guide is educational and does not constitute legal or compliance advice. All figures are stylized illustrations using assumed agent counts, handling rates, pacing ratios, and abandonment rates, and do not describe any particular operation. Abandonment caps, message requirements on abandoned calls, permitted calling hours, attempt and contact frequency limits, consent requirements by channel, obligations regarding reassigned numbers, and record-keeping requirements vary by jurisdiction and change over time. Consult your compliance and legal functions before changing any dialing configuration.