Why Collections Scoring Doesn’t Work And How to Fix It!

Whether you are currently using scoring, have tried and failed to use scoring or are considering using scoring in your collections operation, doing it profitably can be an elusive endeavor. You can hardly turn a page in our industry reading without hearing someone tout the benefits of it. But is scoring really all it is cracked up to be? Are companies really making more money because of it, or are they finding it a giant waste of time and money? We will shed light on this, no matter which group above you are in.

Let’s first speak to those who don’t use scoring or have tried and failed. Are you in this group?

  • Your client mandates you have to contact every account no matter what. So you do, mostly.
  • Your corporate culture is that you believe contacting everyone no matter what, squeezing every drop from any and every account, is the best business practice.
  • You got burned, because your experience changed and now you don’t trust those misguided scores. You called on accounts that had a high likelihood of “collectability” scores, but they performed poorly. Conversely, you found that debtors who had low scores actually yielded some decent collections. You lost faith in the scoring model.
  • You used to use scoring but the results were scattered and you have no way to calibrate a scoring algorithm to the complexion of your collection portfolio. So you reverted to calling everyone simply by the way it’s queued up in your dialer.
  • Your operation is too small to even need analytics.
  • The balances you collect are too small, say for example, on video rentals or library books.
  • You have been in this business a long, long time and you believe you can make your own decisions using MS Excel and intuitive or internal intellectual capital.


Scoring is still a strong choice. Why? Because done correctly, it can be inexpensive effective and a good score will translate to increased profitability which, in turn, means you have a terrific return on investment. If you are apprehensive about it, just learn livescore more about it first. Or, if you have tried and didn’t see results, perhaps you were not doing it correctly?

Knowledge is Power, especially for this industry.

To really truly gain the benefits of scoring, you need to do the necessary upfront homework. In doing so, you ensure the most predictive attributes are captured by your scoring model. In addition to the predictive elements, you also have to factor in the variables around your operating costs because, at the end of the day, the key metric is how much, in total, you collected, and not how many people from whom you collected. This is a typical example and it might be something you experienced in your own operation: A collection company scores a portfolio of accounts using a recovery score from a credit bureau. In looking at the results, the scores show the highest scores are on accounts that have the lowest collection balances. However, the “seat-of-the-pants” (i.e., heuristic) approach of the seasoned collection manager told him that he would collect more from the higher-balance accounts, irrespective of the poor scores they received. The result was the manager was proved right and his cash flow velocity improved dramatically over doing what the scoring told him to do. This is an example of how generic scores don’t account for the operational constraints of the particular individual collection business.

The moral of the story: generic, black-box scores are usually not taking in to consideration the best interest of your business. As a corollary, there is a tremendous benefit to investing in your own analytics and it is a consequence of doing your upfront work.

If it is not personal to your company, it is not as profitable to your company.

Even if you are a small operation and think you cannot afford a custom collections model, or if you think your own data set or database is too small, you actually have some viable models. There are, for example, already collection models out there for most asset classes. One can take a model and use it as a template to tune the model for your own collection portfolio. Once you have the model and deploy it within your operation, you will have gained the sort of analytics capabilities that have typically only been within the grasp of larger organizations, and by using this “semi-custom” technique, will achieve this level of analytics at a far lesser cost.

Leave a Reply

Your email address will not be published. Required fields are marked *