Adding context to list searches
For analysts, being able to refine searches to look for specific criteria or behaviour is critical. We make order and customer centric data available in our dashboard via data lists. In order for analysts to gain better insights into chargebacks, we provided a new dedicated list. And whilst we were at it, we redesigned all the lists (new and old!). We added more contextual information and enhanced how we displayed it - helping our clients get the insights they need quickly.
We made the lists more powerful by adding a dozen more filters to enable analysts to find the information they need quickly. We also made it possible for users to save frequently used filters to minimise time spent doing searches.
It can be hard to get additional context from a list of search results quickly so we included search result summaries help resolve this. We added summaries showing the breakdown of last action, score distribution and chargebacks for the search results. This gives our users some context, instantly. This can be really powerful when used in conjunction with tags to quickly analyse the behaviour of a cohort of customers.
Better explanations for fraud scores
The Ravelin fraud score indicates how likely a customer is to be a fraudster, ranked between 0-100. The higher the score, the more likely that the person trying to transact is a fraudster.
Machine learning scores can leave people confused. How was the score calculated? What was the most important contributing factor to the score? Why was a customer scored one way, and a similar looking customer scored another? The fraud score contribution helps to answer these questions, showing what contributed to the score. To make it even easier to understand, we grouped contributions into identity, orders, payment methods, locations and network.
Highlighting similar customers (it takes a thief to catch a thief)
Fraudsters sometimes sign up for multiple accounts under the same or similar name and email combinations. Fraudsters may also share certain kinds of behaviour. To help analysts identify this, we included an indication of similarity on a customer’s profile.
We show the top ten similar customers by behaviour. This could relate to orders, payment methods, locations or other factors and includes things like type of order and velocity of activity. We also show the top ten similar customers by name or email. This helps analysts identify other customers that look similar to the customer they are currently reviewing - potentially uncovering more fraud.
Adding Network Discovery to Ravelin Connect
Ravelin Connect allows you to see networks within your customer base. This can help analysts when it comes to identifying bad actors that may be connected to each other through things like device, email and phone but who otherwise don’t look like fraudsters.
In order to give our clients as much insight into their customer network as possible, we added a new section to Ravelin Connect called Discover. Discover allows you to see the fastest growing and biggest networks. It also helps you identify the most connected phones, emails, devices, customers and cards. This can help to uncover fraud rings, account takeover and voucher abuse.
Supercharging network investigations in Ravelin Connect
It can be hard to pull apart complicated networks within your customer base during investigations. We can definitely sympathise!
To help our users get the most out of Ravelin Connect, we added a bunch of additional features. Interactive Network Insights let you filter a network view and provide summary breakdowns of last action, top domain providers and score distributions across a network.
The Network Depth slider allows you to filter distance from the central node in a network, showing you how ‘close’ an entity is to something else of interest e.g. a chargeback.
Analysts can also hide unconnected nodes so they can filter out nodes that are only connected to one customer. The idea here was to narrow the focus, making it easier to zero in on direct connections and investigate a network.
The additional features make it easier to understand complicated networks during an investigation - the more at a glance insights we can provide to our users, the better. If the network you’re reviewing is proving too complex, or you want to use the data in different ways, you can also download a list view of the network (another improvement we made!)
At-a-glance fraud highlights of Payments and Orders
To help analysts quickly identify fraudsters, we added order and payment method highlights. Instead of trawling through a customer profile to find disparate bits of information, analysts can look at highlights first before digging deeper if required.
For orders, we flag things like status and velocity. For payment methods, highlights include the number of unique cards, associated card countries, number of card registration attempts and velocity.
Merchant-specific heat maps to highlight trouble spots
Sometimes it can be hard to know if a location is associated with a high level of fraud. Our fraud maps visualise the risk level of a particular location. The fraud heat map represents the relative proportion of fraudulent activity in total past activity for each location. When reviewing a customer, this can provide additional context to analysts.
Making location data easier to digest
Whilst we’re on maps - we made design improvements to how we show users important locations associated with a customer. We added an interactive table underneath the map which allows you to view and filter all the locations tied to a customer. The table also shows you how many times an address has been used - giving an analyst even more insight into a customer’s behaviour.
See the future impact of any rules change
Rules can provide important boundaries for machine learning algorithms and can help to make sure business rules are enforced. However, it can be scary to set up a rule without a clear understanding of what the impact might be. Our rule impact service helps to tackle this as it allows users to test the impact a rule may have on their customers before they implement it.
We added context to our rule impact service to make it easier to assess the impact - users can now see the last action, score distribution and manual review breakdowns for the customers that would be affected by a rule based on a sample.
Merchant specific dashboard experience
Business are all different - some information is relevant to one company but not another. Displaying all information for all businesses without taking this into account can generate a lot of noise. We adjusted our orders tab on customer profiles to display specific information relevant to your business. If you send us tickets, events, hotels and travel information associated with an order, we will show that information. If you’re an e-commerce, taxi or food delivery company, we will hide those sections so you only see what’s relevant to you.
We hope you enjoyed reading this blog on all of our new product launches in the last year. If you’d like to keep updated with our progress, subscribe to our newsletter below to get the latest launches in your inbox.