Building and Refining Rules with the Identity Check Confidence Score

I’ve been at Whitepages Pro for over two and a half years and have seen a lot of innovation during my time. We’ve launched international products, introduced new data elements (including a freight forwarder flag and email-to-name matching), and built new partnerships with industry leading fraud platforms. One of the most exciting enhancements I have seen during my tenure is the new Identity Check Confidence Score that launched in July.

The Identity Check Confidence Score is a single, actionable number that indicates how risky a transaction is. We calculate it by considering the 70+ data elements in Identity Check, along with other proprietary inputs across our network, including identity element velocities, transactional frequencies, and linkage histories. The score ranges from 0 to 500, with higher scores indicating a risk and lower score indicating a good identity.

We introduced our score to make it easier for customers to confidently implement data rules using Whitepages Pro Identity Check.

Many of our customers are on fraud platforms that allow merchants to write rules around certain data elements to automatically accept a transaction, reject a transaction, or trigger it for manual review. In these platforms, an accept rule trumps everything and a reject rule trumps a review rule (in other words: Accept > Reject > Review). For example, one data element we return is whether the name and email from the order match each other. If it doesn’t match, the transaction is more likely to be fraudulent.

While the individual data signals are good indicators of fraudulent transactions, they’re often not strong enough alone to outright reject the order or even put all of those orders into the manual review queue.

Most merchants end up avoiding accept rules altogether because they are just too powerful. Even if 50 review rules fire on an order, a single accept rule will cause that order to be auto-accepted. The result is the writing of even more review rules, causing their manual review queue to get larger and larger.

Our Confidence Score is designed to solve this issue by allowing merchants to fine-tune data element rules. One of the best ways to leverage our Confidence Score is by using it to modify existing review rules. The merchant’s identify rules that are less efficient than they would like and add a condition to make it better.

AVS rules (address verification service) are a great example of this as they typically catch many good customers in the manual review trap. This is what an AVS rule would look like before and after implementing a Confidence Score rule.

  • Before: If AVS = Mismatch; then Review
  • After: If AVS = Mismatch AND Confidence Score > 400; then Review

Chargeback reduction is often a goal of merchants, and now they can easily write a standalone rule around the Confidence Score:

  • If Confidence Score > 450; then Review

Merchants can also use Confidence Score to add more conditions to Accept rules in order to temper their power:

  • If Dollar Amount < $100, AND Bill-to = Ship-to AND Confidence Score < 50; then Accept

By doing this, they not only catch the riskiest orders in manual review but also allow the good customers to be auto-accepted.

At the end of the day, we built the Confidence Score so merchants could find value in our data very quickly. There are some “out-of-the-box” modifications that you can make to your ruleset, but if you want to maximize your ruleset’s effectiveness, please reach out to us. We know how decisions are made on these fraud platforms and are here to help.

You can learn more about the Identity Check Confidence Score here.

Thanks for reading! You might be interested in these posts, too: