Settlement Authority: What Escalation Costs | HL Hunt
Settlement Authority: What Escalation Costs
A consumer on a $1,840 balance offers $1,196. The agent can go to $1,472 and no further, so the account is escalated. 41% of consumers do not come back, which destroys $490 of expected value on a conversation that was already going the right way. The arithmetic says widen the authority, by $86,590 on a 2,000-account portfolio. The right answer is that a single authority was never the correct shape.
What you'll learn
What a tight authority costs
On a $1,840 balance, an agent with authority to discount 20% can settle at $1,472 and no lower. The consumer on the line is willing to pay 65%, which is $1,196, and says so.
The agent cannot accept. The account goes to escalation, which takes a couple of days and requires reaching the consumer again.
- Consumers who disengage before the escalated offer lands: 41%
- Expected value of escalating: 0.59 × $1,196 = $706
- Lost against accepting on the call: $490
The loss is caused by the gap between willingness and authority, not by the size of the discount. Per our early-stage analysis, a consumer who has engaged is in a state that does not persist, and every process step between engagement and resolution has a survival rate attached to it.
And escalation is slower than its stated duration. Per our contactability analysis, the two days to get an answer internally are followed by the problem of reaching someone who was reachable once, which is the part that produces the 41%.
What a wide authority costs
The obvious fix is to widen the authority. That has a price too, and it is real.
If agents can settle at 65% without approval, some consumers who would have accepted 80% are offered 65% instead. Each of those accounts gives up $276.
| Policy | Cost per account in the portfolio |
|---|---|
| Tight authority, 22% of accounts push past it, losing $490 each | $107.88 |
| Wide authority, 30% of the other 78% over-discounted at $276 each | $64.58 |
| Difference | $43.30 in favour of wide |
On a 2,000-account portfolio, that is $86,590 a year in favour of the wider authority.
Which is a finding, not a recommendation. It says that on these parameters the cost of friction exceeds the cost of leakage, and the parameters are the whole argument: the 41% disengagement rate and the 22% push-past rate are the two numbers that decide it. An operation with a 15% disengagement rate and a 40% push-past rate gets the opposite answer, and per our prioritization analysis most operations have never measured either.
The 41% is the number to measure first
Everything in this analysis turns on how many engaged consumers are still reachable after an escalation, and almost nobody tracks it, because the accounts that disengage do not generate an event. They simply do not come back, and they are recorded as unresolved rather than as lost at a specific step. The measurement is straightforward: tag accounts at the moment an escalation is raised, and report the resolution rate for that cohort against accounts resolved on the first contact. Per our cost to collect analysis, this is the same discipline as measuring cost at the margin rather than on average.
Neither answer is the right one
The comparison above is between two flat policies, and a flat policy is the wrong shape for the problem, because the cost of over-discounting is not the same on every account.
Discount authority should be widest where the expected recovery is lowest. An account with a 9% chance of recovering anything loses little by being offered a deep discount, because the alternative was probably nothing. An account with a 70% chance of paying in full loses $276 by being offered the same discount.
| Account profile | Authority | Reasoning |
|---|---|---|
| High recovery probability, recent, engaged | Narrow | Leakage is expensive and the account is unlikely to be lost by waiting |
| Moderate probability, some contact history | Standard | The case where measurement matters most |
| Low probability, aged, hard to reach | Wide | The alternative to a deep discount is usually zero |
| Consumer has proposed a specific figure | Wide, within limits | A stated number is information that an escalation destroys |
| Balance or ownership not fully verified | None. Stop | Per our balance accuracy analysis, settling a wrong figure is worse than not settling |
The last row is a different kind of rule from the others. The first four are commercial judgements about where to place a threshold. The fifth is not negotiable at any expected value, because an agreement on an unverified balance is an agreement that may have to be unwound, and per our furnishing analysis the credit reporting consequences of a settled-for-less entry on a wrong balance outlast the recovery.
Authority set once for every account is wrong for almost all of them
Discount authority should be widest where expected recovery is lowest and narrowest where it is highest. HL Hunt AI Debt Collection scores accounts on expected recovery and sets the agent limit per account, so a willing consumer is not routed through an escalation that loses 41% of them.
Authority is not only about the discount
Discount percentage is the dimension everyone sets. Three others matter as much and are usually left undefined, which is where unauthorized agreements actually come from.
- Term length. An agent allowed a 20% discount but no term beyond six months will refuse an arrangement that pays more in total over twelve. Per our counteroffer analysis, the structure that passes the rule is frequently worse for both parties than one that does not.
- Down payment. Authority to waive or reduce the first payment is a separate dial, and per our re-default analysis it is the one most correlated with whether the arrangement survives.
- First payment date. An arrangement starting on the consumer's next payday holds far better than one starting on the sixth of the month because the sixth is the system default.
- What the agent may say about outcomes. Statements about credit reporting, legal action, or the status an account will be given after settlement are not discount authority, and per our complaint analysis they generate a disproportionate share of complaints.
Item three is the cheapest authority to grant and the most consistently withheld. It costs the operation nothing to let an agent set the first payment date to the consumer's pay cycle, and per our re-default analysis the difference in adherence between arrangements dated to income and arrangements dated to a system default is large enough to dominate the discount question entirely.
Item four is the one that creates liability rather than cost. An agent who says the wrong thing about what will be reported has made a representation the operation has to stand behind, and widening discount authority without constraining representations widens both at once.
Making wide authority safe
The objection to wide authority is not really the arithmetic. It is that discretion is harder to supervise, and that objection is answerable.
- Set the limit per account before the call, so the agent is working inside a boundary rather than exercising judgement about the boundary.
- Require the consumer's stated figure to be recorded, which converts the deepest discounts into a documented response rather than an agent decision.
- Confirm every agreement in writing before the first payment, with the amount, the schedule and the account status on completion.
- Sample the agreements near the authority limit, not a random sample. Per our monitoring analysis, a random sample of a rare event finds nothing.
- Report leakage as a measured number, meaning accounts settled deeper than their score justified, so the cost of the wide policy is visible rather than feared.
- Review the disengagement rate quarterly, since it is the parameter the whole policy rests on.
- Keep the verification stop absolute and separate from everything above.
Item two does most of the work. An agent recording that the consumer offered $1,196 and accepting within authority is in a very different position from an agent choosing to offer $1,196, both for supervision and for the file, and it also removes the main incentive problem with wide authority: an agent cannot anchor low if the low number has to come from the consumer first.
Item five is what makes the policy reviewable. Per our benchmarking analysis, an operation that cannot say what its leakage is will set authority by anxiety rather than by evidence, and anxiety consistently sets it too tight, because the cost of leakage is visible on individual accounts and the cost of friction is spread across accounts nobody connects to the policy.
The deepest discounts should be documented responses, not agent decisions
Recording the consumer stated figure before accepting it changes both the supervision problem and the anchoring problem. HL Hunt AI Debt Collection captures the offer, applies the per-account limit, and confirms the agreement in writing before the first payment.
Frequently asked questions
About $490 of expected value per account, because 41% of engaged consumers disengage before the escalated offer reaches them.
On these figures wide wins by $43.30 an account. But a lower disengagement rate and a higher push-past rate reverse it, so measure both.
No. Widest where expected recovery is lowest, narrowest where it is highest, and none at all where the balance is unverified.
Term length, down payment, first payment date, and what agents may state about outcomes. The first payment date is free and matters most.
Key takeaways
- Escalating a willing consumer destroys about $490 of expected value because 41% disengage.
- The loss comes from the gap between willingness and authority, not from the size of the discount.
- Tight authority costs $107.88 an account; wide authority leaks $64.58, favouring wide by $43.30.
- On 2,000 accounts that is $86,590, but a lower disengagement rate reverses the conclusion.
- The disengagement rate after escalation is the parameter everything rests on and is almost never measured.
- Authority should be widest where expected recovery is lowest and narrowest where it is highest.
- Where the balance or ownership is unverified there should be no authority at any expected value.
- First payment date is the cheapest authority to grant and affects adherence more than the discount does.
This guide is educational and does not constitute financial, legal, or compliance advice. All figures are stylized illustrations using assumed balances, discount levels, consumer reservation prices, disengagement rates, and portfolio sizes, and do not describe any particular operation or portfolio. Permissible settlement practices, required disclosures, restrictions on statements about credit reporting and legal action, written confirmation requirements, and the tax and credit reporting treatment of settled accounts are governed by federal and state debt collection law and vary by jurisdiction and account type; consult qualified counsel and compliance advisers before changing settlement policy.