Earlier last month Amazon sent shockwaves in the fraud prevention industry when they announced the release of the Machine Learning based Fraud Detection AWS Service. Amazon Web Services (AWS) is Amazon’s Cloud Infrastructure and is widely used by many of the largest internet companies such as Netflix, Slack, Adobe and even NASA. It’s by far the largest cloud infrastructure services and one of Amazon’s crown jewels in terms of profitability.
When it comes to eCommerce fraud prevention Amazon is also a leader in technology and innovation. Simply given the sheer volume of ecommerce transactions that Amazon processes each year – just to put things in perspective, Amazon’s eCommerce revenue was $203 Billion in 2018 alone – being at the forefront of fraud detection technology isn’t an option it’s a must. Even a 0.5% reduction in fraudulent transactions will amount to more than a billion dollars worth of revenue.
As such it was interesting to see Amazon announce the availability of machine learning based fraud detection service architecture for AWS. It’s unclear if the target audience of this service are merchants who are already using AWS, third party developers or payment processors as there are no use-case examples on the AWS website yet and the service seems to be in ‘Preview Mode’ for now.
What we do know is that users need to provide source data such as emails, IP addresses and other historical transaction and account registration data, along with signals which indicate the fraudulent transactions and which are legitimate transactions. Amazon uses this data against its algorithms — along with fraud algorithms developed on the consumer business of Amazon’s business — to train customized models that recognize things like potentially malicious domains and IP address formation. After the model is trained, users can create, view, and update rules to enable actions based on model predictions. In more technical terms The Fraud Detection AWS Service page overview deck states:
“Fraud Detection Using Machine Learning enables you to execute automated transaction processing on an example dataset or your own dataset. The included ML model detects potentially fraudulent activity and flags that activity for review. The diagram below presents the architecture you can automatically deploy using the solution’s implementation guide and accompanying AWS CloudFormation template.”
The Fraud Detector Services lets users introduce additional rules based on risk tolerances. For example, they can set up a customer account registration workflow to require additional email and phone verification steps only for registrations that display high-risk characteristics. Furthermore, Fraud Detector can flag accounts that are more likely to abuse ‘try before you buy’ programs and flag suspicious online payment transactions before orders are processed and fulfilled.
It’s all exposed through a private endpoint API, which can be incorporated into services and apps on the client side. Amazon claims that Fraud Detector’s machine learning models identify up to 80% more potential bad actors than traditional methods, on average.
Impact on the eCommerce Fraud Prevention Industry
So should dedicated fraud solutions be concerned with Amazon’s foray into this space? In my opinion most likely not. Many merchants are already weary of Amazon as it is. After all Amazon accounts for almost 50% of all US based eCommerce transactions and using Amazon’s infrastructure to process sensitive transactional data makes it a tough pill to swallow. Some merchants such as Wal-Mart have officially stated that they will flat out refuse to work with any vendors who use AWS based architecture for their solution.
Amazon’s fraud solution however may be attractive for midsize niche merchants already using AWS and that have the internal resources to build out a fraud service in house since most of the heavy algorithm lifting is already built by Amazon. In addition, third-party developers who may want to build a fraud solution from the ground up or payment processors who wish to add another layer of machine learning fraud detection functionality to their service may find the solution appealing.
It will be interesting to see how companies leverage this service in the future, I don’t think any of the bigger and more established and enterprise focused dedicated fraud solutions will have anything to worry about. However, some of the smaller players who focus on SMB merchants may start to feel a more competitive landscape.
I believe we’ll see a rise in fraud solutions catering to various niche markets and the industry will likely get a bit more crowded in the next few years. The revenue model for such solutions is still very lucrative and there are still massive opportunities available. Needless to say the eCommerce Fraud industry will be an exciting space in the coming years.
If you want to dig more into the AWS Fraud Detector Service and take a look under-the-hood you can read the Implementation Guide. Please keep in mind this is a very technical document.