Esusu helps renters build credit by reporting on-time rent payments to credit bureaus. I was brought in to answer the next question: now that we have this rental data, what else can we do with it? The answer was mortgages.
Esusu's core insight was simple: millions of Americans pay rent on time every month and get zero credit for it. Credit scores are built on credit cards, auto loans, and mortgages — none of which renters necessarily have. So Esusu built a platform that reports rental payments to Equifax, Experian, and TransUnion, giving tenants a credit history they could not get anywhere else.
They sell this two ways. Directly to tenants who sign up themselves, and to property management companies whose building owners want tenants to have an incentive to pay on time. Phase 1 was working. Esusu had a growing dataset of verified, on-time rental payment histories across thousands of tenants.
Phase 2 was my assignment: these tenants eventually want to buy a home. Can we use this rental history data to actually help them get a mortgage?
Traditional mortgage underwriting relies heavily on credit score, which is exactly what Esusu's users are trying to build. A tenant who has paid $2,400 in rent on time every month for three years is demonstrably creditworthy, but a conventional underwriting model might still reject them because their FICO score is thin.
Fannie Mae and Freddie Mac, the two entities that back most US mortgages, had recently started allowing rental payment history as a supplemental data point in underwriting decisions. The door was open. The question was whether Esusu could walk through it.
Early in the project I mapped out two distinct approaches to getting rental data into mortgage underwriting decisions. Both had merit and different tradeoffs on speed, data quality, and lender adoption.
I recommended pursuing both in parallel, with Path A as the near-term pilot to establish lender relationships, and Path B as the longer-term product that would make Esusu's underwriting tool available to any mortgage applicant in the country. The roadmap I built reflected this phasing.
Path B required building a verification layer that could look at a bank statement and confirm: did this person pay rent on time, every month, for the past 12 to 24 months? Plaid's transaction API could pull this data with user consent. The challenge was turning raw transaction data into something a lender could trust.
I worked with engineering to define the data schema and wrote the requirements for how the system should classify transactions as rent payments, handle irregular amounts, and flag months where verification was inconclusive. Getting this right mattered because lenders would be making credit decisions based on it.
The compliance and legal constraints were the most time-consuming part. Every data handling decision touched fair lending law, and the requirements that came out of those conversations shaped what we could and could not build.
Building the product was only half the job. The harder half was getting lending partners to agree to use rental history data in their underwriting decisions. Lenders are conservative by design. Anything outside the standard FICO-based model is a risk they have to justify to their own compliance teams.
By the end of the internship we had moved several lenders from initial conversation to signed pilot agreements. The 22% improvement in pilot-to-contract conversion came directly from addressing these objections with better data and cleaner integration docs.