Blog Startup & Dev


E-commerce fraud prevention using a monthly statistical model

In order to prevent e-commerce fraud, you need to be aware that frauds has their own characteristics due to their nature. Indeed, fraudsters are not standard buyer like everyone else because of their goals which are different from a standard buyer. Those “cyber-thieves” would have a result-oriented behaviors when normal people would have a regular behavior.

Studied population

These data has been collected from our client using our Anti-Fraud module. The study includes 57,732 orders placed between July 2014 and March 2015 on e-commerces. The french market has been chosen for this seasonality analysis for the fact of its unique time zone that eases understanding of the analysis.

E-commerce fraud rate steady throughout the month

Many hollywood movies about bank heists and frauds make us believe that there should be a logical fraud pattern during a month. From our observations, we see a slight increase between the first two weeks and the last two weeks with a fraud rate which go from 0.9% to 1.4% which is over 50% more fraud.

E-commerce fraud prevention: fraud rate during the monthFraud rate (by day over the month)

However, this upward trend of fraud rate over the month is attributable to the volume of daily total purchases that decreases throughout the month as the fraud rate remains stable.

E-commerce fraud prevention: number of orders VS number of fraudsOrders volume VS Frauds volume (by day over the month)

Our study focuses on the French e-commerces. As a result, the downward trend in the purchasing behavior is mostly due to the monthly salary cycle in France. By nature, frauds are not subject to such salary constraints.

Conclusion on the monthly distribution of e-commerce fraud

In this study, it seems there is no correlation between the orders volume and frauds volume. In this case, no specific fraud pattern appears to be revealed despite the particular topology of study area (France): the number of frauds is stable while the numbers of orders is not. Therefore, in terms of e-commerce fraud prevention, a corollary would be that:

The later an order is placed in a month, the greater is the probability of fraud.

Tags: , , , , , ,

Leave a Reply

Your email address will not be published.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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>