Build vs. Buy: 5 Capabilities You Need in a Fraud Tool
By Kevin Lee /
18 Jun 2018
When considering whether to build or buy your fraud prevention tools, there are a lot of criteria to assess. What is your in-house level of technical expertise? What are your time constraints? Would developing this technology be a core competitive advantage? And more…
In our recent webinar, Building versus Buying: Understanding the Right time for Each Option as You Grow, we outline all of the most important questions you need to ask during this important decision. We also run through the five capabilities that will make up your “full stack” for fraud prevention.
Whether you will build or buy, aim for a solution that includes all five of these layers:
Let’s start from the bottom…
Layer 1: Data
The data you collect for fraud prevention will come from in-house. And that’s great! There’s so much rich information that your customers are leaving on your app – from their device information and IP location to the pages they navigate to and the buttons they click on – that your fraud solution will have plenty of data to learn from.
Is homegrown data the only option? Actually, no…for example, if you use Sift Science, you also benefit from access to data from across our global network of customers. That means that anytime a user takes action on another customer’s site or app, their risk score automatically updates.
Layer 2: Machine learning
Speaking of which…now that you have that data, what are you going to do with it? Machine learning is the most scalable and flexible technology for proactively preventing fraud attacks before they happen. It continually learns from new data – in Sift Science’s case, our Live Machine Learning gathers new intel from across our entire customer network, and updates immediately.
If you’re on the fence about whether machine learning is a necessary part of your fraud stack just think: fraudsters are already using it themselves. These days, fraud networks are moving beyond brute force attacks. Instead, they’re using sophisticated technology like machine learning to try and reverse engineer your systems. You need to fight fire with fire, and lean forward with machine learning.
Layer 3: Workflow and rules automation
Some companies think you need to have either rules or machine learning – but you can have both! While machine learning is a “smarter” approach to stopping sophisticated fraud, you probably still have a need to enforce certain policies, like which countries you can and can’t ship to. You don’t need machine learning for this — you can just use rules in an automated fashion to make sure it doesn’t happen.
Layer 4: Analyst tools and feedback
If you’re using rules and workflows, you’re going to need a robust console or other tool to manage them. Talking with companies, we’ve learned that teams often spend more than 50% of their time just collecting data by logging into multiple tools — rather than spending time on more strategic tasks, like analyzing that data or making decisions. The average merchant uses 9+ tools to fight fraud.
A robust fraud tool can consolidate all of that data into one place, so you don’t have to search for it. Remember that analysts are probably your most expensive line item. They need to be able to make decisions quickly and efficiently.
Layer 5: Reporting
The final piece of your fraud stack? A way to measure your success. You’ll need to track your KPIs and how well you’re meeting your objectives. Having a clean, clear reporting structure in place will help you make better decisions going forward.
Want a deeper dive into whether you should build or buy your fraud solution? Watch our free webinar!
Kevin is the Trust and Safety Architect at Sift Science. Building high-performing teams and systems to combat malicious behavior are what drive him. Prior to Sift, Kevin worked as a manager at Facebook, Square, and Google in various risk, spam, and trust and safety roles.