A Strategy To Deal With Credit Card Chargebacks (From a Data Analyst PoV)

Creada con caryion.com

By Miguel Ángel Díaz| Analytics Team

Dealing with credit card payments chargebacks is tough and scary. So we bring a very condensed strategy sketch to try to help to come up with some solutions. The following diagram is explained through this article.

Merqueo’s Strategy diagram against Credit Card Chargebacks.

First thing is to define when Chargebacks are a problem, this is the very first step, and it’s shown on the upper side of the diagram. Merqueo’s approach was to set a 2% limit on the payments with chargebacks / total of payments (chargeback rate), this limit is the same some payments platforms recommend in order to work with them. It’s very important to keep track (keep asking on diagram) of this metric since suddenly fraudsters can ‘attack’ or some other problem can appear, this is a very dynamic problem. That’s why even when the answer to the question ‘Are chargebacks a problem?’ is no, you should always Keep asking.

If the answer to this first question is Yes then it’s highly recommended to make an Exploratory Data Analysis. It can be tempting to try to find 3rd party tools to help the problem, but It’s not recommended, since chargebacks are not always related to fraud. That’s why it’s recommended to make an Exploratory data analysis over the operation, payments system, orders and clients. This step can be very helpful since the problem/s could be isolated. For example, it may be found that there is a problem with the payments system, and clients are being double charged. Or that all the fraud payments come from new accounts, certain IPs, zones or/and even email domains. 

Next step is Try deploying a set of rules and/or analytic models whose objective is to detect or deny possible fraud payments. Trying to find a solution here will force you to understand the problem at your business. Merqueo’s operation it’s almost unique and that’s why it’s very important to try to deploy an inhouse solution, since no other 3rd party tool will be as adapted. In the worst of the case you could try to deploy a model that helps to identify the most ‘suspicious’ payments .

In case you don’t get to find a reliable solution, It may be worth trying 2 things, gathering more information in order to retake the process from the exploratory data analysis or/and implement a Manual review. Gathering more information with the objective of finding some relevant information. Manual Review means that a person reviews all the suspicious payments. This could also be a good complement to any analytical solution. 

After getting more information it’s always valuable to answer the question: Does a 3rd party tool help? Maybe tools could provide more information or solutions that help the developed solutions. But it is at this point where a 3rd party solution could be valuable, after most have been done to understand the problem.

After these steps it’s very likely that a clear idea of how the solution would look for your business, then it’s time to Implement Solution.  Again this is a very dynamic problem, a solution that works today couldn’t in the future, so remember, keep asking.

Following this strategy and after several iterations the solutions was:

  •  A set of rules for Colombia (It wasn’t possible to deploy a model due to lack of data). For Brazil and México  an analytic model and a set of rules for each country.
  •  Manual review for the critical cases in all the countries
  •  A 3rd party tool was contracted,  and is used to provide models and rules with more information about clients emails, ips and phone numbers.

With this strategy Merqueo achieved a 20% reduction of the chargeback rate, saving more than 50 000 US in Mexico and keeping the chargeback rate most of the time under the setted limit.