Plaid is known for providing its customers with trustworthy, high-fidelity bank, payroll, and document income data, which can be used to make smarter approval and loan term decisions.
When connecting to someone’s payroll or bank account isn’t an option, documents are a tried method of income data, and Plaid needed a solution to help determine the accuracy and trustworthiness of those documents.
Document fraud is a huge area of concern for lenders. In research Plaid conducted with over 400 lending leaders, 61% say they have experienced document fraud — making it the most significant type of fraud experienced.
“The primary goal of our Credit products at Plaid is to allow lenders to seamlessly collect their borrowers’ income and assets to verify borrowers and grant differentiated loan terms using bank account data,” said Rohan Sriram, Product Manager at Plaid. “It’s really important for our customers to be able to gather data with high fidelity.”
When it comes to income verification, Plaid offers three different ways for customers to prove their income. The first two (payroll income verification and bank income verification) allow the loan applicant to connect their payroll or bank account directly with the Plaid platform as part of the lender’s onboarding process. The third option is through document review, which is when an applicant uploads documents or images that are then analyzed to assess the person’s income.
“When information is coming straight from a payroll account or a bank account, there's a high level of confidence that the direct from source data is legit and can be directly used for underwriting,” explained Rohan. “But when it comes to document or image uploads, there's a little bit of a question as to the accuracy of the information provided. Did the applicant create a fake document using a template? Did they take a valid document and change the name or alter transactions? How trustworthy is this document and the information it contains?”
Finding answers to those questions through painstaking, meticulous manual reviews could take hours or even days. What’s more? There’s no guarantee that the decision the analyst makes is the best one for the company or the customer.
“The main pain point that we're trying to solve is ensuring the document data we give our customers has very high fidelity. Ultimately, they want to trust the information that they have and use it to make the right decision.”
Rohan Sriram • Product Manager, Plaid
To address this need, the team at Plaid set out to find a partner that would help them quickly and accurately assess document and image uploads while identifying cases of document fraud as part of the customer onboarding process.
Since each underwriter’s risk tolerance can vary, the Plaid team also wanted to make sure that they gave clear context to their customers by providing detail on why a document was rejected, as well as the overall trustworthiness of the file.
Enter Katie Randolph. As Business & Corporate Development lead for Credit Products at Plaid, she worked side-by-side with Rohan to narrow down the vendor pool through network referrals and independent research.. Together they short-listed three vendors – one of which was Inscribe – and set out evaluating them based on five distinct criteria:
1. Latency: Could the vendor deliver near real-time results?
2. Accuracy: What is the scope of the tool’s detection capabilities and how precise is it in identifying instances of fraud?
3. Document variety: Could the tool support image uploads or screenshots via smartphone, in addition to traditional documents, like PDFs?
4. Pricing: How flexible is the price structure and what value for money does it represent?
5. Lending use cases: Does the vendor have experience working with a lending company in a similar capacity?
“Ideally, we wanted a partner that checked all the boxes and understood the high-stakes nature of the problem our customers need to solve,” explained Katie.
Plaid ultimately found that partner in Inscribe. They were particularly impressed by Inscribe’s 30-second processing time, as well as the accuracy of the product in detecting fraud on both documents and images.
“About 40% of our submissions are image-based files, so image review is important to us and our customers,” noted Katie. “That’s not something that a lot of vendors offer, so it’s a point of differentiation in the market.”
Before making their decision, Katie and Rohan worked with Inscribe to set up a comprehensive pilot program. As part of the process, they compared Inscribe to another automated document review tool, as well as a customer who relies on manual checks as a control case. Ultimately, Inscribe came out on top.
“There was a lot of rigorous analysis that went into the selection. I was particularly impressed at the ‘all hands on deck’ attitude from Inscribe during the pilot and the level of attention and support we got. The fraud detection capabilities were obvious and we were able to witness that first-hand.”
Katie Randolph • Business & Corporate Development lead for Credit Products, Plaid
Today, Inscribe is a central element of the customer onboarding process for lenders with document-based processes who rely on Plaid’s Document Income product.
When a person who is applying for a loan or mortgage reaches the income verification stage, they are prompted with options on how they want to provide income verification. If the applicant chooses to upload documents, Plaid uses the Inscribe software to analyze the files for any signs of tampering or fraud. Inscribe then provides the customer a trust score for each document, which is a numerical indication of how trustworthy a document is. Inscribe also reviews any risk signals found in the document, which Plaid can then share with their lending customers for further context.
To produce these risk signals, Inscribe models are trained on hundreds of millions of data points to detect signs of digital tampering, evidence of forgery, and the use of templates purchased online. The tool also produces a clear explanation and visual of exactly what’s been altered in a document, whether it's a name change, an adjustment to a bank statement balance and transaction, or both. Users can cross references the difference between the document provided and the document recovered and even drill into the metadata to see if any software’s been used on the document.
“Inscribe has this ability to look at what was actually manipulated in the document,” said Rohan. “The X-ray feature has been really helpful for our customers. We can take them under the hood and actually show that the name has been tampered with or that someone has cleverly put a layer on top of the pay stub. These are things that the human eye would likely miss so it’s pretty cool to see how the technology works.”
Plaid is only at the beginning of their customer journey with Inscribe and has already provided several pieces of feedback for how they’d like to work with the product team to expand use to meet certain business needs.
“Broadly speaking the reception has been really positive,” said Katie. “The implementation was fairly straightforward, and we were able to stand up the solution quickly. Plus, the Inscribe team has been receptive to our feedback. Overall, we've had a great experience and would recommend working with Inscribe.”
For Plaid’s customers, there are two main benefits to Plaid partnering with Inscribe: Time savings and workforce optimization.
“During the pilot we saw the potential to automate 50% or more of applicants through the process. The fact that Inscribe is able to come back with a response on each document in under 30 seconds has really helped our customers move applications a lot faster.”
Rohan Sriram • Product Manager, Plaid