Why embedded payments programs stall inside banks
Banks keep announcing embedded payments and then quietly shelving them. The one I launched inside a top-20 bank got to $10M in nine months. The difference was three decisions made before the first line of code.
Every large bank has an embedded payments slide. Most of them have had it for three years. Very few have a business.
I ran one that worked: an embedded payments and merchant services business launched inside a top-20 bank, through a fintech the bank had acquired. It produced $10M of income in its first nine months and served 22,100 businesses in year one. Here is what was different, and it had very little to do with technology.
The program was sold to software companies, not merchants
A bank's merchant services team sells to merchants. It has done so for forty years, through branches and referral partners, and it is good at it. Embedded payments is a different sale. The customer is a software vendor who serves a thousand merchants, and what they are buying is not processing. They are buying a revenue line, a boarding experience their users will tolerate, and a partner who will not embarrass them in front of their customers.
We built what we called the Integrated Referral model specifically for ISVs, and we started in one vertical, healthcare, where the software vendors already knew their merchants' pain. The economics made the case on their own: about $9,000 of lifetime value per merchant against $452 to acquire one. That ratio is why the business got funded for year two.
The programs that stall try to sell embedded payments with the merchant sales team and the merchant pitch. The ISV hears "we would like you to refer your customers to us" and reads it correctly as "we would like your customers."
The risk model covered every relationship type, up front
This is the decision the bank's second line cares about, and it is the one most programs leave until a deal is on the table.
There are six ways a bank can be in a merchant relationship: Referral, Retail ISO, Wholesale ISO, ISV or Integrated Referral, PayFac Rent, and PayFac Own. Each one puts a different amount of underwriting, monitoring and liability on the bank. If your program only has a policy for the first two, every ISV conversation ends at the point where the ISV asks for the fourth or fifth, and your compliance team says "we will need to look at that."
We wrote the risk and compliance strategy for all six before we signed the first partner. It took longer. It meant that when a partner wanted a PayFac-style arrangement, the answer was a document, not a committee.
Somebody owned the P&L, and that somebody was on the sales calls
An embedded payments program inside a bank usually reports into a product group, or an innovation group, or a partnerships group. It has a budget and a roadmap and no revenue target that anyone will be fired over.
Ours had a P&L, and I owned it, along with sales, marketing and operations. For the first deals I was on the calls myself. Not because the sales team couldn't do it, but because in the first six months the product changes every week, and the person who can change the product needs to hear the objections unfiltered.
The other thing owning the P&L does is force the boring work: bundles, a boarding experience that did not lose half the applicants, reporting the ISV could show their own customers. None of that is exciting. All of it lifted conversion.
What this means if you are starting one
If you are a bank starting an embedded payments program, I'd check three things before any technology decision.
Who is the buyer, and does your sales motion match them? If the answer is "merchants, through our existing team," you have a referral program, not an embedded one.
Is there a written policy for all six relationship types? If not, your first serious ISV will stall in compliance, and the second one will hear about it.
Who owns the number? If the person owning the product is not owning the business, you will be a feature factory, not a growing, successful business.
Get those three right and the technology is a vendor decision. Get them wrong and the technology will not matter.