SaaS API founders should not miss this week's Cartoon Avatars interview with William Hockey, former Plaid cofounder/CTO and now founder of Column.

He does not do interviews often and rarely do you get this level of insight into a $13 billion, fintech category-defining behemoth. What follows is a TL;DR for those who, well, TL;DW.

This article assumes Plaid needs no introduction.

The $12 Billion UI Decision - Owning the UI

Many SaaS API providers take pride in being "behind the scenes", or being "whitelabel" to appeal to as many enterprise customers as possible. Jeff Lawson often proudly talks about how many people use Twilio without realizing it.

TLDR: Hockey bet the company on going against that received wisdom - forcing 100% of his customers to migrate to a Plaid hosted UI with Plaid's logo and branding - over a period of 2 years, with a lot of pushback.

He estimates this decision alone was worth 90% of the company value today (!)

From the 22-26 minute mark:

I think the most successful decision we made was actually owning the interface - the physical design and owning the client side...

When we first started, we were this transparent infrastructure provider and so the consumer had no idea who we were...

...and so what happened is you didn't know, as a consumer, that Plaid existed in that flow. We realized that this was kind of problematic because, as a consumer, you were not getting the same experience hooking up your bank account to Venmo as you were when you hooked up your account to Square Cash, or Chime or, Coinbase..

and that had a lot of security issues but also had conversion issues because every application thought that their design was best or whatnot...

so what we decided to do is we made them display a Plaid designed UI to the consumer...

We made the application insert our branding, our logo and our experience into the application. That was extraordinarily controversial, as you can imagine, because these applications want to control the experience.

One way to view this move is concluding "Plaid customers were so bad at their jobs of optimizing UI that just providing the APIs wasn't enough."

The motivations were two fold: self protection (Plaid oriented) and conversion optimization (customer oriented):

We needed to establish some level of relationship with the consumer and provide uniformity across these applications because we were the only one focused this hard on conversion.

It actually started converting a lot better... the consumer actually started to feel comfortable, like hey I know this screen, I've seen this before, and it also allowed us to do a lot of micro optimizations around messaging certain banks and just allowed us to kind of have a platform that we could actually deliver content and software directly to the consumer...

Where my eyes really popped is how far they took this - forcing ALL their users to adopt this flow - near impossible for most API companies to do especially if customers threaten to leave over this decision.

Now 100% of traffic flows this way and it's actually one of the only reasons that we have good relationships with the banks because those sensitive data never actually hits the application anymore and we can also if a bank wants to make you accept some terms of service or something like that, we can deploy that instantly... and so it allows us instant flexibility.

but it was a very very challenging rollout - it took almost 18 to 24 months, there was a lot of pushback to it - but i think if we didn't do that, A) consumers wouldn't have as good of an experience, B) we also would have got commoditized and it would have been really easy for these applications to switch it out. It would just been a worse experience for everybody

But in the end, it was worth it:

I think that (decision) probably generated like 90% of our market cap today.

You can try the full UX of the $12 Billion UI right here: plaid.com/demo without connecting a bank account.

You can see Stripe, a Plaid competitor that stayed relatively behind the scenes, increasingly start to own the experience with Billing in 2018, Checkout in 2020, and lots more I am unaware of. In fact, Stripe Checkout's marketing sounds eerily similar to what Hockey just said for Plaid:

You get the benefit of all this and everything that’s to come: even faster load times, additional payment methods we add, compliance with future payments regulations, and every optimization we make to maximize conversion—all without major code changes on your end.

Alvar Lagerlof also reports that Swedish fintech Klarna also inserts a branded UI:

This is a product direction you can expect more SaaS APIs taking going forward as they seek both to build their own customer relationships and to serve their B2B customers better (by doing their jobs better than they can).

Sidenote: Founder Intuition Over Data

What is perhaps most interesting is that this decision was made without data - Hockey felt like he would not have had support from consumers, banks, or employees - which is why these kind of high conviction bets require founder-led companies.

Also noteworthy - it took "three or four years" before it became obvious that forcing their UI was the right decision.

At 27 minutes:

I see this with founders a lot - when they go try to do product interviews or customer interviews, they're assuming that the people they're interviewing have similar knowledge, interest, or insight to them, and that's just not the case.

We just felt like there was a bit of an arbitrage where we knew where the industry was going to play out we knew what the banks were going to react and so we just made the gut call.

This isn't something he encourages at all scales - and the transition from being a product visionary at 10 employees to a delegating leader at 1000 employees is a difficult but necessary transition.

Column: The Bank with Developer Experience

For his next act (as a billionaire, post Plaid), William bought a bank and is now aiming to provide "financial infrastructure to other companies" - if anyone wants to do bank-like activities (loaning, holding, or moving money), Column offers you an API to do that. They aren't the first, but they are aiming to do it better/faster than the incumbents.

True to form, he seems to want to own the whole experience here too.

At the 39-44 minute mark:

How it worked before is, you had this very complicated supply chain where you had these community banks that were obsessively renting out their charters, but they didn't really know what people were doing with it.

(Then) banking as a service providers... would sit in between the customer and the bank and they would provide the APIs, they'd provide the developer experience, and they would pass all the risk and complexity up to the bank...

and the bank would then outsource all of their technology out to a core and these cores are something called FIS, Fiserv and Jack Henry and they provide all of what you actually think a bank does...

We looked at this space and it seems over complicated for something that should be relatively simple... so we are a bank but we also have built all of our own core, we built all of our own developer experience, so if you want to do something, all you have to do is come to us and we'll provide the end-to-end solution.

We took this really elongated supply chain and condensed it into one institution...

What i learned from Plaid is it's really hard to build a really great developer experience unless you control your entire end-to-end flow because if I don't actually own the bank or I don't own the core, I can't actually deliver you a good service, because I have partners upstream that could it mess up.

What happens in the industry right now, is you have these really great companies that are super smart, they build really great technology, but they've obsessively offloaded a lot of the technology and responsibility upstream to people they don't control and so they actually can't deliver a good experience.

I want to control the entire thing, all the way up to the Fed and back...

So in many ways, it's just a high conviction bet that I think this space is going to evolve a completely vertical solution that is controlled end to end, it will be the winner and I think it's going to take me 10-40 years to get there, but if we do I think we have an opportunity to be the largest and most important financial institution out there.

The rest of the conversation goes into cryptoskeptic discussion which isn't super useful for devtool founders.

Founder/Investor Reflections

This interview/anecdote was notable because it concerns the ideal structure of developer tooling (I wrote a basic intro on Horizontal vs Vertical strategy here).

The default approach of all SaaS APIs for the past 1-2 decades has been to aggressively horizontalize - pick one piece of functionality that is repeatedly built by all other companies, specialize in it, turn the fixed-cost upfront investment into a scale-from-zero-to-infinity variable cost API.

Hockey took a different route - going from partial vertical integration - forcing Plaid as an API-in-the-middle provider - to full vertical integration with Column. Other founders, notably Rippling, are choosing to centralize rather than specialize, and build multiple offerings from day 1.

Developers know the adage - there are only two ways to make money - bundling and unbundling.

Economists call this the study of industrial organization.

As exhaustion sets in from the modern data, devops, martech, and other landscapes becoming too fragmented, there are surely opportunities to offer the vertically integrated, premium "Apple" to the free, commoditized, so-so "Android" in each kind of developer tooling domain.

Either way you land, Plaid and William Hockey's story is a fascinating and relatively rare example of successful vertical organization that more people should know.