Blog / machine learning
Five new product updates from the past few months
We’ve been working on changes that can help Fraud Managers and Fraud Analysts to work more effectively and make their lives easier. Here’s a round-up from the past few months...
The Ravelin team has had a very busy few months! We’ve been continuously collecting feedback from our users to help us improve our products.
We know Fraud Managers have a lot on their plates - they are typically responsible for the team conducting manual reviews, managing the chargeback process, reviewing refund decisions, monitoring and updating fraud tools, fraud reporting and meeting fraud KPIs.
As well as overseeing their team and making sure they’re happy, they also need to set fraud strategies, liaise with the business leadership and work closely with other departments. There’s a lot to do! Therefore, we’ve been working on changes that can help Fraud Managers and Fraud Analysts to work more effectively and make their lives easier. Here’s a round-up from the past few months.
More oversight on team activity
To help give Fraud Managers better oversight on team activity, we’ve introduced the Activity feed. This section of the dashboard lets you see all activity for individual team members relating to things like manual reviews, comments, annotations, rules and tags. You can read more about reviewing team activity in the Ravelin Help Center.
You can now also see the manual review history on a specific customer profile which can be useful when reviewing team decisions over time if multiple decisions have been applied.
Reporting and analytics improvements
Reporting is a key aspect of a Fraud Manager’s role - fraud reports are often shared with the wider business and the leadership team. We’re constantly making improvements to our Analytics to support regular reporting and save Fraud Managers time.
We’ve made several updates recently, including:
- Adding the ability to select multiple segments for analysis, for example market country, currency and last action source when creating reports
- Monitoring individual team members’ manual review performance over time, segmenting and filtering by count, email and review label
- Reviewing when sources disagree with each other, for example when a fraud score disagrees with a manual review or rule via additional reporting on action sources
Improvements to rules
Rules can be a useful tool alongside machine learning because they can be used to set boundaries for specific business policies your company may have. Based on client feedback, we’ve redesigned and improved our Rules functionality.
The new designs make it easier for you to review rules, especially more complicated rules with multiple conditions. You can easily see the real impact of each rule, looking at the cases where the rule was directly responsible for the action in place. You can also check rule performance in Analytics.
As part of recent rules work, we’ve made more rule conditions available including things like customer ID, order IP address country, latest eWallet type, order item categories and order value (USD). It is now much easier to create or edit a rule; you can filter by condition types and every condition shows a description and an example.
Allowing you more flexibility to express doubt
Making definitive decisions on customers can be very tricky - having some flexibility in the kinds of actions you can take on an individual customer is really important.
Fraud Analysts may want to give customers the benefit of the doubt in some cases, and having only binary good/bad labels can make this hard. We’ve added the ability to forgive a specific chargeback - This gives you more flexibility in dealing with customers who have a chargeback but are not fraudsters.
In addition, we’ve added two additional options when reviewing a customer: Potentially Fraudster and Potentially Genuine. You can configure the underlying logic for both, allowing you to choose a setup that works best for your existing processes, risk appetite and customer base.
Making voucher abuse thresholds more flexible
For clients using our Voucher Abuse API, it is now possible to set thresholds based on the count of voucher users and the depth of the network. This can be configured differently depending on the voucher type. This makes the Voucher tool more flexible across different kinds of marketing campaigns.
There are loads of other exciting improvements we’ve made to the dashboard in the last few months but these are a few of our favourites. If you have any feedback or questions, please get in touch.