Fintech Economics — Digit

Disclaimer: I don’t believe I know anyone at the main company, digit.co, discussed below.

We’re seeing some maturity coming to the Fintech market recently. 3 “big” changes, SoFi upped its wealth fees for non-borrowers, Betterment raised its fees, and the big one trending online as I write this is Digit converting from free to paid. I’m going to focus on this last announcement below.

At a high level this makes sense for Digit, it’s them doubling down on their main product, before focusing on “cross-buy” (as the banks are currently calling it post-WF fiasco), but I think it deserves a bit more inspection about the economics they’re seeing today and why they made the changes.

Assumptions

Some basic “fun facts” first:

  • Publicly have announced they’ve moved about 500MM in USD over company history. Let’s say 2/3 is still on file at their bank partner.
  • Cash deposits of this scale make around 110BPS for the company, but that was before 2 rate increase.
  • Digit gives back annually 20BPS to their customers, prior to fee plans ($2.99 / monthly with 100 day trial).
  • Average active Digit customer saves monthly between $80 and $170.
  • Have also seen claims of CAC of ~$5 based on a few online sources (one good twitter thread). This is likely based upon a $5 referral bonus.

Note: None of this data comes from Digit directly, but there are only so many variables to manipulate, so the maths here isn’t too complicated.

Maths

If my assumptions hold true, top line revenue appears to come to $222K monthly — biggest question is savings retention.

$500MM* 2/3 = $333MM * 0.0080 / 12 = $222K monthly revenue

Looking deeper at the assumptions, and assuming a spherical cow, let’s look at a few different savings scenarios for Digit customers. From this it’s relatively easy to deduce a few internal metrics.

This models 110BPS yield for deposits, which is accidentally different than above.

Based on the above assumptions, I feel its safe to say that average balances in Digit are lower than $4,000, if not much lower, to justify this change. Plus with a CAC of ~$5 their payback period likely ranges from 4–6 months based just on revenue.

So how does this change unit economics for Digit? I built two barebones models to lay out the two different businesses based on above assumptions, and while clearly the fee based model makes much more money, I think there were better ways to have gone about this.

Savings Based Model

Fee Based Model

See spreadsheet here for details: https://docs.google.com/spreadsheets/d/1iP-Igirf6iifPwTmPSfz0tqDfrkx0LmJ1tbvBlBZKTw/edit#gid=1837601610

At a high level, purely based on numbers, a fee based customer looks much better. That being said, I’m being pretty lenient with attrition numbers, and this can quickly flip a profitable customer to unprofitable if retention numbers are off, and that is excluding that this likely greatly increases CAC beyond the 100 day trial.

What do I think they should do?

This is where this post gets fun, given that I don’t have to run their business and might have some fundamentally flawed assumptions above. If I were them, and could stomach it, I’d:

  • Match savings rate within 10BPS to market top:(http://www.bankrate.com/banking/savings/rates/)
  • Don’t charge any fee, ever for Digit basic. Free is much different than even $1 in customer’s eyes.
  • Continue with $5 referrals to continue low CAC growth.
  • Focus only on growth to get largest deposit base possible.

And then either, flip it to a bank looking for deposits at ~1% cost basis, or get into cross-sell of other products, most likely lending given cheap deposit base. Whether you like it or not, if you’re in retail financial services you’re going to look like a traditional player given enough time or scale.

That’s it for now, will update this post as people call me out on being wrong on things.