Data Show Software Payfacs Are Contracting. Here’s Why

Payment facilitators, or payfacs, must be registered directly with the card schemes. This provides interesting data to analyze as it relates to their location and growth.

Recently compiled data and visualizations of payfac trends point to some interesting trends.

There are a few things that stick out to us that we should discuss.

First is the churn in payfacs.

North America and the EU both offer payfac models, but the EU is different in that payfacs must also have a banking (PI or EMI) license. Banking licenses obviously come with more cost, so many European companies prefer to go the payfac-lite approach, which is basically operating as a subprocessor on a payfac.

What’s interesting is the high amount of churn in the payfac model (called un reg. for unregistered in the following graphic):

If you look at North America on the left, there’s a lot of churn in payfacs. 52 of 257 – or 20% – unregistered in 2020. Then as fintech got hot in late 2020 and 2021, the number of payfacs expanded only to again see more churn heading into 2023 as fintech markets cooled.

Why is this?

Payfacs are sexy… and misunderstood.

We don’t think a company should be a payfac at less than $2B of processing volume.

And even that isn’t a hard rule on the lower bound.

We know there will be people who hiss at our comment, but the math just doesn’t pencil.

Processors dropped their pants and gave amazing deals to the first wave of payfacs in the early 2010’s (Toast is now at a reported $0.004 buy rate, for instance).

That’s pretty cheap.

The cost of writing directly to the card schemes – at scale, mind you – is $0.001 per auth and settle (meaning a roundtrip cost is 2 tenths of a penny).

Then you need a sponsoring bank which is going to take some additional cost (or a direct acquiring license, which is also costly).

But is it worth spending the millions of dollars and years of engineering + approval time to go from $0.004 to an effective $0.002?

Toast has said it isn’t.

Same for Square and other large payfacs that have over 11 figures of processing volume, because processors can and have given large payfacs pricing at what amounts to half a basis point of cost.

Today, payfacs aren’t getting these rates. We’ve seen payfacs doing single digit billions in volume only to have effective buy rates of $0.05 once you factor in all their nonsense monthly fees… and they’re servicing merchants with $20 average checks.

That’s an effective 25 bps, which is f*cking expensive.

We don’t think it makes sense for most software companies to be payfacs.

And we suspect the payfac attrition rates to be a consequence of software companies realizing that the payfac model was uneconomical.

Our thinking is only further supported with the following graphic:

Note that ISV representation as a percentage of overall payfacs is shrinking in the US.

Instead, payfac growth is coming from non-ISV entities that are likely using the payfac model for payments disbursements.

Here’s an example.

McDonald’s wants to collect their franchise fees from the payment stream across all its franchised locations. McDonald’s becomes a payfac, orchestrating all the funding and payouts across all its franchisees. In this way McDonald’s can automatically collect franchise fees from the franchisees’ payments stream, and then pay franchisees through the same mechanism should the need arise.

We think that the market has already figured out that it’s reached peak software payfac, which is why you see the growth coming from payfac-lite plaforms offered by the likes of Adyen, Infinicept, Stripe, Tilled, and others.

In this setup these named entities (Adyen, et al.) operate the master payfac and software companies become white labeled partners underneath the payfac. The named entities hold the contract with the merchants of the software partner and also manage the flow of funds.

The payfac lite model allows software companies to get stood up quickly with ~90% of the perceived value of a payfac without the additional costs and timelines associated with actually running the payfac.

The more interesting trend to us, however, is one that’s not yet being picked up:

Embedded payments, while convenient, are creating de facto monopolies and increasing the total cost of ownership for merchants.

Absent payments choice the merchant is faced with high switching costs and, as a consequence, pretty shitty support (why improve customer service if you know the merchant can’t do anything about it?).

This is something worth watching, especially as many of these software companies need to show rule-of-40 growth and are finding it so. much. easier to add bps to their payment fees over building, selling, and supporting value-add features (that merchants don’t and won’t pay for anyway).

There IS a software solution to this problem, breaking merchants from mandated, embedded processing and allowing merchants (and ISOs) to choose whatever processor they want, regardless of what their software tells them they can use for payment processing.

But it’s still in private beta.

Things are going to get very fun over the next two years.