Shared API for Fragmented Payroll Systems
Jeremy Zhang, CEO of Finch, on building a universal API for employment systems
The real bottleneck in digital 401(k)s is not selling the plan, it is wiring into whatever payroll system each employer already uses. Human Interest could justify building 40 to 50 connectors itself because every payroll run has to pull employee census data, update deductions, and move money into retirement accounts, but even that only reached about 60% of customers because HR and payroll are far more fragmented than most software markets.
-
In Finch's framing, the top 10 payroll and HR systems cover only about 55% of the market, and getting closer to 75% coverage would require 40 to 50 connectors. That means each new employer sold by a benefits company can still bring a brand new integration problem.
-
What falls outside connector coverage does not disappear, it turns into labor. Human Interest described a workflow of staff manually setting data, adding deductions, pulling CSVs, and standardizing payroll outputs, which shows why integration work becomes an operations headcount problem, not just an engineering task.
-
This is why payroll partnerships matter so much for 401(k) providers like Human Interest and Guideline. Both depend on clean pay stub level data to run compliance, handle reversals, and keep plans in sync, and Human Interest later expanded to more than 400 payroll integrations as it scaled past 20,000 businesses and 1M participants.
The next phase of this market is a shift from in house connector teams to shared infrastructure layers. As more payroll, HR, and embedded payroll products fragment the system further, the winning retirement and benefits companies will be the ones that treat connectivity like core infrastructure, then spend their product energy on onboarding, compliance, and employer experience instead.