Approval rates follow a curve every risk manager recognizes. A strong model and an expert data science team can carry it up, and then it flattens.
When we launched Plaid Guaranteed Payments in May, the promise was straightforward: send us a transaction, get a decision in milliseconds, and if we approve it, we guarantee settlement. Teams saw their approval rates rise and got back to building their products.
But we knew there was more to do to improve approval rates incrementally based on:
What you already know about your user. An account history with some blemishes may belong to your VIP.
The pace of fraud. Vectors are shifting faster, attacks are increasingly sophisticated, and point-in-time checks are no longer sufficient.
The shape of the decision. Approve-or-decline decisions on the full transaction amount lack flexibility on transactions where there is partial confidence.
Today's release addresses all three opportunities. Over the last quarter, Guaranteed Payments customers saw dollar approval rates rise more than 22%, compared to baseline.
Custom approval treatment for your most trusted users
You know things about your users that a broader network wouldn’t be able to infer from their bank accounts: data points like user tenure, wallet balance, and loyalty tier. An account that looks risky across the Plaid Network may belong to a user with a six-month history with you. Without your tenure and tier data, our models can only approximate that distinction.
This matters most at the top of your user base, where VIP tiers can get their own approval treatment. Custom attributes help Plaid tailor decisions for the users you already trust.
Fraud doesn't stop at sign-up
Many fraud programs front-load their scrutiny. You verify at onboarding, then trust the account. But account takeover, mule activity, and device farming all target established accounts precisely because that's where the reviews drop off.
Guaranteed Payments now refreshes additional device, account, and behavior signals for every transaction, alongside existing abnormality detection. We capture updated device intelligence and user metadata at the time of payment to connect the dots between fraud rings and a single user.
The result: risk that emerges after sign-up gets flagged before the payment is initiated, and a payment from a returning user on a known device can be more confidently approved.
More configurability, higher approval rates
Binary approve or decline decisions remove nuance from the experience. We now support more variations of coverage to increase dollars approved.
Partial amount guarantees: When Plaid can confidently back a portion of a transaction, we will recommend that you offer that amount to your user. If the user confirms, we’ll then guarantee that portion instead of declining the entire payment, giving you and your user flexibility to decide how to handle the remainder.
Deferred guarantee: For certain high-risk transactions, Plaid can still guarantee the full payment amount. Coverage begins after a short observation window.
Built on the Plaid Network
None of this works without the data underneath it, or the team that builds on it. The model behind Guaranteed Payments is developed and continuously tuned by the data team that has run Plaid Signal in production for the past five years. Guaranteed Payments runs on Signal, our AI-powered ACH risk platform trained on $311B in transactions, and Protect, our fraud intelligence system drawing from interactions of 1 billion devices across 500M+ linked accounts.
A program designed around your metrics
Guaranteed Payments is a managed product, but it isn't a black box you integrate against and hope for the best. Our data science team builds and continuously fine-tunes the model behind your program, and we design that program with you to meet your product and commercial needs.
Before launch, we configure the rule set and model around your user base, your transaction profile, and the approval rate your business needs. Then we monitor against those targets and tune continuously, segmenting by user tenure, adjusting thresholds on high-dollar transactions, and testing retry timing.
And if a payment does fail, recovery is a partnership. Customer-led recovery lets you keep the direct relationship with your user while Plaid works to complete the payment.
Get started
To sign up for Guaranteed Payments, contact your Plaid account team or reach out to us here.