Why do a collection agency’s results swing from month to month, and how do you compare one month with another fairly?

The short answer

In our experience most month-to-month swings in an agency’s results come from causes other than a change in how well the floor works. In our experience agencies depend on the placements their clients send, and some batches are good and some are bad. In our view the economy and the time of year change how consumers feel and what they can pay. Who was on the floor and plain chance add to it.

That makes one month hard to compare with the next. Before you decide the floor did better or worse, ask what changed in the paper, in the economy and in the calendar, and compare like accounts with like.

What changes from one month to the next

• The placements you are sent. In our experience no two batches are alike. A client sends fresher accounts one month and older ones the next, or a new portfolio with a different mix. The floor works what it is given.
• How old the debt is. In our experience older debt, and debt that has already been through other agencies, pays less often. A batch that is older than the last one can pull results down without anyone working less well.
• How big the balances are. In our experience a change in the typical balance changes both how often consumers pay and how much each payment is. Larger payments in collections.
• How the accounts score. If the batch arrives with a score, in our view a lower-scoring batch should be expected to pay less. Getting more from collection scores.
• Interest rates and the economy. In our experience a change in interest rates, or in how people feel about the economy, can change how consumers respond very quickly. In our view, when household budgets tighten, the same kind of account tends to pay less often.
• The time of year. In our experience holidays and the weeks before Christmas can move results in a way that repeats from one year to the next, and in our view many US agencies see payments pick up in tax refund season. A December compared with a November can look like a fall that is really the calendar.
• Who was on the floor. In our experience leavers, new starters, holidays and schedules change which collectors work which accounts. Agent matching when collector turnover is high.
• Small numbers. A small floor working one kind of account may see only a handful of payments a week, and that number moves a lot by chance alone.

The mix trap

An agency’s overall rate can fall even when every kind of account pays as well as before. Suppose fresh accounts pay well and old ones pay poorly. If this month’s placements are mostly old accounts, the overall rate falls, even if the floor did a little better on both.

A well-known example is not from collections. P. J. Bickel, E. A. Hammel and J. W. O’Connell looked at graduate admissions at the University of California, Berkeley, for fall 1973 (Science, 1975). The overall figures showed what they called a clear but misleading pattern of bias against women. Department by department, few showed a significant difference, and about as many appeared to favour women as men. The overall gap came from women applying more often to departments that were harder for anyone to get into.

On a collection floor, the departments are the kinds of account. A total that mixes them can tell the wrong story.

How to compare months fairly

1. Keep the definitions still. Make sure payments, accounts and contacts were counted the same way in both months, from the same systems.
2. Read each placement at the same age. In our view a placement’s money arrives over months, so compare this placement’s second month with the last placement’s second month, not one calendar month with another. Liquidation, read by placement month.
3. Compare like accounts with like. Group accounts by what in our experience affects how they pay: age of the debt, balance, how many agencies have worked it, the client and any score. Compare each group with the same group last month.
4. Split reaching people from what happens on the call. Fewer right-party contacts and fewer payments per contact are different problems. More RPCs or better conversion?
5. Ask what changed outside the floor. Interest rates, the economy, the season, a new client rule, a change in contact limits.
6. Test changes against a control group. Run any change on the floor against a control group in the same weeks, so the paper and the economy hit both sides equally. How to prove a routing change worked.

Explaining a bad month

In our experience clients and owners usually see one number. In our view the most useful thing an operations lead can do is show what sits under it:

• Results by kind of account and age since placement, this month against last, so it is clear whether the floor did worse or the paper was different.
• What changed in the placements: older debt, smaller or larger balances, more accounts already worked by other agencies, lower scores.
• What changed outside the floor, such as interest rates or the season, and the same month last year where the season matters.
• What you are changing, and how you will know, measured against a control group run in the same weeks.

The same thinking applies to collectors. A collector’s month swings with the accounts they were given, which is why they should be judged on accounts like theirs. Is it the collector or the accounts?

Where agent matching fits

Agent matching can’t change the placements, the economy or the calendar. It changes which collector works each account. BestPair sends each account to the eligible collector who has done best with accounts like it, from the floor’s own call and payment history, inside the rules and workload limits the agency sets, and measures the result against a control group. Because both groups work the same placements in the same weeks, a good or bad batch can’t pass for a gain, or hide one.

Sources

• Sex Bias in Graduate Admissions: Data from Berkeley: P. J. Bickel, E. A. Hammel and J. W. O’Connell, Science 187, 1975; the classic example of an overall rate that misleads when the mix changes.