An important distinction in banking-as-a-service (BaaS) is between the product itself and the way that the product is consumed. https://twitter.com/julieverhage/status/1340742782751805442

Fintech startups in BaaS are, universally, awesome at making their services easy to consume.

Great APIs are table stakes in this world.

Credit-data-as-a-service. Fraud-prevention-as-a-service. Compliance-as-a-service. And so on...

BaaS APIs are critical for consumer-facing fintech startups because they are developer oriented, they want to move fast, and they don’t want to waste resources on things that won’t differentiate them.

But the product underneath the APIs matters a lot more.

A lot of BaaS products aren’t actually that differentiated.

Integrating data from centralized providers for credit, fraud, or KYC and applying rules isn’t new.

There are many many legacy vendors that already do it and probably do it better and at a larger scale.

These BaaS startups win with B2C fintech startups because of the ease of working with their APIs (and, to a lesser extent, because they swim in the same fintech VC waters).

But as those B2C fintech startups grow, they may want a solution from a legacy vendor that’s more scalable or sophisticated. Or they may want to erase some technical debt by building it themselves.

And the BaaS startup can’t pivot and sell to incumbent banks because their differentiating trait (ease of integration) doesn’t matter if you already have a legacy solution that works.

(FWIW: all incumbent banks have legacy infrastructure solutions in place)

The BaaS infrastructure providers that I’m bullish on bring a differentiated product (usually wrapped in a beautiful API) to the market.

Something that delivers new value to B2C fintech startups *and* incumbent banks.

Something other companies can’t easily build.

This is why I like Plaid and MX and Finicity and all the new companies trying to aggregate payroll data.

Building out a network of bank/payroll provider integrations — building coverage where none existed — is highly differentiated and not easy to duplicate.

It’s also why I like companies that are truly innovating in the fraud management space.

Building fundamentally new technologies like ID scanning, behavioral biometrics, or Self-sovereign identity solutions.

These BaaS startups take a lot longer to take off because the use cases they enable are new.

There’s no pent-up demand for them.

But once they take off, they’re