A process: optimise acceptance rate

David Wilde

The acceptance rate is a very important metric, for our clients as well as for us. It is hotly debated, measured and interpreted differently. One thing is clear: the higher the acceptance rate, the happier our customers are, because it means more sales for them.

Customer, buyer and Ratepay “control” the acceptance rate

What is also clear: in the high-risk business of invoice, direct debit and instalment payments, an acceptance rate of 100% is unrealistic. But of course we always try to get the highest acceptance rate for the merchant or marketplace. Many factors play a role in this. Some Ratepay can control, others not. One factor we cannot control is which buyers use our payment methods. These are largely what we call “typical people”. However, there are also the people who cannot afford the items because they are in private insolvency proceedings, for example. On top of that, there is fraud (mainly identity theft for us), a very ugly side of e-commerce.
Factors that we can control are mainly technical. This also includes the qualified handling of our risk appetite.

Risk management in the light of a sustainable and flexible service

This means that in order to provide a sustainable and innovative service to our customers, we have to be careful not to overstretch ourselves when it comes to buyer risk. In summary, we divide shoppers’ ordering needs into three categories:
a) Typical buyers,
b) buyers who, in our opinion, overstretch their financial limits, and finally
c) people or even machines who pretend to be buyers but want to cheat us and the dealer.
In the case of typical buyers, we aim to allow all order requests. In categories b and c, we have to weigh up the risk we take. We are particularly attentive to fraud here.

The more customers we have and the more diverse they are, the more each individual customer benefits

We have already collected a lot of behavioural data across many sectors in terms of order preferences. Among other things, we base our initial risk profiles for new customers on this. For example, we start with a certain acceptance rate for each new retailer. As soon as a trader goes live with us, his trafik is under close observation. From this, we draw conclusions in short iterations to further optimise the entire system. This means:
– We identify technical errors that may not have been noticed during integration,
– We now observe real live graphics of the trader and can thus optimise the risk settings to fit exactly.

A stairway to the optimal acceptance rate

What does the optimisation look like? First and foremost, we look at the ordering behaviour. The focus is on a rapid reduction of the false-positive rate. For us, false positive means that we have rejected an order that would have been paid for. Over time, we also get a better understanding, for example, of fraud patterns, which in turn allows us to further adapt our prevention system.
Over time, this process looks like a dance: step one is to reduce the false positive rate, step two is to increase fraud prevention (the statisticians among you can also use the false negative rate). In this way, depending on the context, we achieve a kind of optimal acceptance rate within a few months. This dance is all the more challenging the more diverse the customers and products of the trader or marketplace are.

In this way, we have developed an agile, lean and closely aligned adoption rate improvement cycle for our clients.