The Fraud Prevention Space’s Complex Optimization Problem

The Fraud Prevention Space’s Complex Optimization Problem

This is the first post in a five-part series, written to support you in the process of choosing the fraud prevention solution most suitable for your business and its needs.

A matter of priorities

The amount of resources and effort eCommerce brands invest in leading potential customers through their purchasing funnel to checkout is enormous. Shoppers can drop off at so many places along the way. They can decide to ignore the campaign and not visit your website altogether. If they’re already at your website, they can leave their cart empty. And if they added something to the cart, they can just abandon it and decide not to decide, or even worse – go and purchase the chosen product at a competitor’s website. 

It might seem like you are about to read a marketing article, but happily, this is not the case. In this series of blog posts, we want to focus on the one aspect most eCommerce players overlook in their effort to grow revenue, increase customer lifetime value (LTV), and provide the best customer experience to the potential customers who already decided to become their customers – Fraud Prevention Optimization.

Fraud prevention optimization is a matter comprised of two variables: 

  • The cost of fraud – to keep things simple we will call it “chargeback costs”
  • Approval rates – the number of incoming transactions, minus the number of transactions that are declined due to fraud, or suspected fraud

After speaking with hundreds of merchants, we came to the conclusion that most merchants are focused on keeping the cost of fraud down, as far as possible from the 1% (now 0.9%) that will get them on the Visa and MasterCard excessive chargeback programs. They do so, even if it means compromising approval rates. They focus on one variable instead of solving this optimization problem by trying to reach a healthy balance. 

Another mistake most merchants make is to look at this optimization problem in an aggregated manner. Meaning, looking at the overall chargeback rate and approval rate and not breaking it down to markets, product offering, etc. Let’s take an example to better understand this point: 

Here is some information about an online merchant and relevant industry benchmarks: 

  • Overall annual sales volume – $1B
  • Approval rate – 95%
  • Chargeback rate – 0.2% 
  • Industry average: approval rate 93% ; chargeback rate: 0.3% 

What do you think about this merchant state from a fraud prevention perspective? Has this merchant successfully solved the optimization problem we presented above? Let’s look at some additional information that will assist in answering this question: 

  • US Market 
    • Annual sales – $400M
    • Approval rate – 97%
    • Chargeback rate – 0.15% 
  • EU Market 
    • Annual sales – 400M
    • Approval rate – 95%
    • Chargeback rate – 0.25%
  • Asia Market
    • Annual sales – 200M
    • Approval rate – 91%
    • Chargeback rate – 0.45%

What would you say now? Is this optimization problem solved successfully?It’s hard to tell; maybe the Asian market deals with much higher rates of fraud and this is what leads to a higher chargeback rate even when the approval rate is lower. Or maybe, this team is less experienced in coping with fraud in the Asian market, for different reasons, and it could actually improve the performance for this market significantly? 

Let’s explore the immediate opportunity this merchant has. Every 1% increase in the approval rate in the Asian market translates into $2M of increase in sales (which is much more than just monetary value but we will get to it later). Also, maybe it is possible to keep the fraud cost for this market lower. Every reduction of 0.1% in the fraud cost translates into $200K in spendings. 

It seems like a pretty significant opportunity, doesn’t it? If you were this merchant’s CEO, what would you instruct your team to do? 

You’d probably instruct the team to explore ways to optimize performance in the Asian market. The team would have two options for doing so: 

  1. Try to optimize elements in the current solution to improve performance
  2. Evaluate other fraud prevention vendors who can assist with this goal

Over the next four blog posts we will break down the process of choosing a new fraud vendor into three steps: 

  • Step 1– Noting all the major benefits of partnering with a new fraud vendor
  • Step 2 – Breaking down the different solution types that are out there – rule-based, scoring, machine learning, chargeback guarantee, and listing the pros and cons of each
  • Step 3 – Lastly we’ll cover the key considerations you should have in mind when choosing a fraud prevention partner  

Note: In the above example we simplified the scenario and broke performance down into only three markets. In reality, especially for larger eCommerce merchants, it is wise to measure performance for each market (by IP or bin), each product offering, 3DS segment (if you are using it) and many other variables in order to reach full optimization. In next week’s blog post, we’ll provide a deeper explanation of how to breakdown your sub-segments to optimize performance.