What counts as success when measuring agent matching in collections?

The score: dollars collected per account contacted

In live pairing (Stage 2), the accounts in scope are split at random: 80% are allocated by BestPair and 20% stay on your current routing as the control group. Each month both groups are scored on dollars collected per account contacted. The extra is how much more the BestPair group collects per account contacted, multiplied by the number of BestPair group accounts contacted. That is the measure in the Order Form, and the fee follows it.

Three reasons it is per account contacted:

• The groups differ in size. The BestPair group holds four times as many accounts as the control group, so totals cannot be compared. Per account contacted puts them on the same footing.
• Contact is where the collector matters. BestPair changes only which agent works an account. It does not contact consumers, and it does not change whether an account is worked, what a consumer is offered, or any collection terms. Once an account is contacted, who has the conversation is what differs between the groups.
• It weighs amounts as well as counts. A floor can raise how often contacts pay and still collect less, if the payments get smaller. Dollars per account contacted shows both at once.

Both groups run in the same weeks, on the same mix of accounts, with the same collectors, so the season and the placement mix hit both alike. How a fair routing test works.

Why a promise to pay is not the score

A promise to pay is an outcome code your collector enters at the end of a call. A payment is money in your collection system, and it often lands days or weeks later. Some promises are kept in full, some in part, and some not at all. A routing change that raised promises without raising payments would add nothing to what your agency or its clients receive.

So the score is built from the payments file: payment date and amount, keyed to account ID. Reversals come in as negative amounts, so a payment that bounces or is reversed is taken back out. Your outcome codes still matter. The outcome codes list flags which codes are right-party contacts and which are promises to pay, so both groups are counted by your floor’s own definitions.

RPC conversion as an earlier read

RPC conversion is the share of right-party contacts that end in a payment or an arrangement. What RPC conversion is, and how to measure it. It is known when the call ends, which makes it a useful early read on whether conversations in the BestPair group are going differently.

It does not weigh amounts. A small instalment and a large settlement each count as one conversion, and a gain can sit entirely in the size of what is collected. In the published case study’s longer read of the live test at a North American collection agency (one telecom book), the share of reached accounts that paid was about the same in both halves, and the difference came from the larger accounts. The case study shows how that was counted. Watch conversion for movement. Judge on dollars.

Settle the definitions before the start

Agreeing what counts before the first account is split keeps the result from becoming an argument later. Most of it is already fixed:

1. The score. Dollars collected per account contacted, both groups, scored monthly, as the Order Form sets it.
2. Your outcome codes. Which codes count as a right-party contact and which as a promise to pay, written once in the outcome codes list.
3. The campaigns in scope. Listed on the Order Form, so both groups come from the same paper.
4. What result would count for you. Agree it internally before the start, and read it on full months with adherence at or above the 70% minimum.

Year-on-year comparisons, a single strong week and a rise in promises are all things a floor sees during a test. None of them is the score. BestPair does not guarantee a level of lift, and the monthly comparison is there to show a flat result as plainly as a gain. Why the first weeks can trail.