- Blog
- The psychological economy of DPD 60, and why “most people…
Collections Strategy
The psychological economy of DPD 60, and why “most people pay on time” stops working
The same sentence carries different weight as an account ages. Here’s how to choose the right message for a customer who is two payments behind.
Picture a message that has worked for two years. It goes out to every account a few days past due, and it ends with a single line: most of our customers pay on time. Reply rates are good. The team adds it to the day-30 sequence, and it performs there too.
So they add it at day 60, where recovery is hardest, and any lift would be worth having. Reply rates fall. Over the following weeks, complaint transcripts start carrying a word nobody expected: judged.
The line itself is accurate. What changed is who receives it. A customer who has already missed two payments knows they are not among the people who pay on time, so a message organized around that comparison tells them something they have already worked out for themselves and offers nothing they can act on.
This is the part of the psychological economy of a late payment that a fixed schedule cannot see. The same sentence carries different weight depending on where the reader believes they stand, and that position moves as an account ages. Two large field experiments show how far apart the results can land.
Social proof is the strongest lever in one portfolio and useless in another
Social proof rests on a finding that has held up across decades of research: people decide what to do partly by reading what others around them appear to be doing. A late payment is a good candidate, because most customers have no sense of what normal looks like in that situation.
Researchers tested exactly this with 7,063 late-paying clients of a Colombian bank. Their randomized trial found that any message at all reduced the likelihood of being late by 4%. Researchers tested several message types against each other, and only the social norms version reached statistical significance on its own. More striking, its effect was still measurable 14 weeks later, making it the most durable approach in the set.
A second trial points the other way. Among 9,196 people holding unpaid hospital bills, researchers compared appeals to social norms, appeals to the public good, and simple personalization. Neither of the first two produced a significant effect. Addressing the recipient by name did.
Two properly randomized studies, opposite conclusions. The populations differ in country, debt type, and relationship with the creditor, so neither result can be lifted into another portfolio without testing it there first.
A norm message only works while the customer still counts themselves in the group
Every social norm message makes one quiet claim about its reader: you are one of these people. Everything depends on whether they agree, and agreement erodes as an account ages.
-
Early on, the message reminds someone they belong. At day 5, "most customers pay by the due date" describes a group the customer is still part of, and the message reads as an invitation back.
-
Later, the same words land as a verdict. By day 60, after two missed payments, the line says the customer is outside the group. Being the exception motivates nobody, and few people pay a bill in order to rejoin a club they have just been told they left.
-
It can also make staying behind feel normal. If a customer's own frame of reference is other people in arrears, a norm about on-time payers has nothing to do with them, and a message hinting that many customers are behind quietly confirms that position.
What changes between 30 and 60 days is the question the customer is answering. At 30 days, it is still arithmetic: how much is owed, how much is available, and where this bill ranks against rent and groceries. That is the ground psychology in economics covers.
By day 60, there's a second question on top, and it's one about identity. The person who's missed twice has already started to see themselves as a certain type of payer. A message that positions them in comparison to the reliable payers is precisely the kind of message that will resonate with that self-view. That's why the same comparison works as a nudge at day 30 and as a form of validation at day 60.
The design rule is: name a group that the customer can still envisage themselves in. 'Most people pay on time' and 'most customers in your situation set up a plan' are both forms of social proof, and only the second one is still applicable to someone who is two payments behind.
The right message depends on which group the customer can still claim
The rule above produces three usable approaches, and the right one depends on how far the account has drifted from the group a message can credibly name.
-
A membership norm compares the customer to people who pay on time, and works while that group is still plausibly theirs.
-
A situational norm compares them instead to people in the same difficulty, which stays credible much later because it names where they actually are.
-
A no-norm message drops comparison altogether and personalizes through the name and the amount, which suits customers for whom any group framing is likely to land badly.
Each example below carries a compliance note, because the wording of a collections message is constrained by what a lender can actually substantiate and deliver.
Day 5, membership norm
Hi [Name], most customers clear this before it affects anything. Your $82 payment is 5 days late. Here's the link to settle it in one tap.
The customer can still see themselves in that group, so the comparison works. Check that the claim is actually true for this portfolio, and don't hint at consequences that won't happen.
Day 60, situational norm
Hi [Name], most people who've fallen two payments behind sort this with a plan rather than one payment. Here's a plan starting at $25. Reply YES, and it's set up.
The comparison is to people in the same position, and it names what they do about it. Don't suggest a plan is guaranteed if eligibility rules apply.
Day 60, no norm at all
Hi [Name], your account is 60 days past due. The smallest step that helps today is $25. Reply PLAN, and we'll set the rest up.
No comparison at all. The name and the amount do the personalizing, so identity never comes up. State the status accurately and claim nothing about other customers.
Social proof only helps customers who could pay
The third row deserves more attention than it usually gets, because disengagement is rarely a decision anyone makes consciously. Ignoring a message costs less effort than opening one, and the mind defaults toward conserving effort wherever it can. In cognitive economy psychology, that default is the baseline rather than the exception, and once the habit of skipping a message forms, each unopened one makes the next easier to leave alone. The way out is changing what the message says rather than how often it arrives.
Deciding which customer gets which message is a daily, per-account judgment
Norms work on customers who still identify with the group. Knowing that doesn't tell you which of your customers those are, and the answer shifts from week to week across a portfolio of any size.
That's the problem AI agents are built for. Rather than running a fixed sequence, they decide per account who to contact, when, on which channel, and with what message, then adapt as each customer responds. Norm-based messages go to the customers they still suit, and the rest get something matched to their own situation.
Whether that works depends entirely on what the agent is told to optimize for, which is the whole subject of AI agent evaluation in collections. An agent evaluated on opens rather than completed payments will get steadily better at the wrong thing.
The decisions also improve only if the system learns fast enough to matter. A multi-armed bandit moves volume toward what performs while the test is still running, instead of holding a losing message in market until a significance threshold clears. Both sit inside a broader move toward digital collections built on behavior rather than a calendar.
Testing at that scale reaches variables a human program would never get to. In one Tier 1 telecom deployment, KredosAi ran more than 4,400 message variants in market, varying channel, time of day, language, and even font color to find what actually drove payment rather than opens.
Research can give you the rule. Only your own portfolio can show you where the line falls inside it.
FAQ
What is social proof in collections?
A message telling the customer what people like them do, such as "most customers pay by the due date," intended to make payment feel like the ordinary choice rather than an unusual one.
At what point does a 'norm' message start to backfire?
When the customer no longer counts as part of the group that's named. At DPD 60 (after two missed payments), "most people pay on time" becomes a judgment rather than a nudge, and it's for that reason that reply rates and complaint language tend to move together at that point.
What should replace social proof at DPD 60?
Personalization and a concrete next step. In the hospital billing trial, using the patient's name improved payment rates in cases where a standard appeal had failed. A situational norm that names the individual in the same position falls somewhere between the two.
Is norm-based messaging a compliance risk?
The framing isn't the issue. The issue is accuracy. Any statement about what other customers do has to be true in relation to that portfolio, and any implied outcome has to be consistent with the actual process of the lender.
What is cognitive economy in psychology?
The mind's tendency to conserve mental effort, relying on shortcuts and defaults instead of deliberate reasoning. In collections, it explains why an unopened message makes the next one easier to leave unopened, since doing nothing takes less effort than deciding.
What is the psychology economy of a late payment?
The set of mental trade-offs a borrower makes when deciding whether to pay: which bills feel most urgent, how much effort each decision costs, and how they see their own position relative to other customers. That last element shifts between 30 and 60 days past due, which is why a message that worked earlier can fail later.
```