A Painful $26.5M Lesson
Six years ago, I had the opportunity to lead the engineering team building the reconciliation and operations backend for a broker-dealer. The founders envisioned Motif Investing, a $125M VC-backed retail trading platform for the masses; the Robinhood of our time.
In late 2012, after raising over $25M and 2½ years of development, we finally got the beta out the door. It would be months before we would publicly launch. While a startup faces many challenges, I do not believe Motif was ever able to recover from the enormous costs to get started. I learned just how costly and slow it is to build FinTech applications.
Robinhood was founded almost three years after Motif was founded in April 2013 and raised $3M to build a retail trading platform.
Eighteen months later, Robinhood launched its iOS trading application. While an astonishing accomplishment (with one-eighth the capital and about three-fifth the time as Motif), it was nonetheless a very costly endeavor relative to other successful ventures.
Case in point: In August 2009, Uber raised $200K. In July 2010, Uber released its first product with one-fifteenth the capital and three-fifth the time of Robinhood.
I’m not content with the excuse that it just takes this long to build financial applications. I know we can do better because this is a story we’re already familiar with.
Look to AWS and Firebase for Inspiration
In the 90’s, starting a company required a meaty CapEx budget for servers, an equally meaty budget for systems and network engineers, and months to build out the server infrastructure. In 2000, Rackspace proved that you don’t need a large CapEx budget and with smaller team of system administrators, you can maintain your servers. In 2010, AWS proved that you can spin up a secure, HADR, compliant, CI/CD-enabled infrastructure with no DevOps engineers, a paltry startup budget, and a couple of days of work. AWS proved that it can accelerate software development by at least 2 orders of magnitude and reduce cost by at least 3 while ensuring better quality, security, compliance, and scalability.
Firebase took it even further for mobile development by providing a unified platform with built-in authentication, analytics, testing infrastructure, app messaging, monitoring, and much more.
There is a clear macro trend in software development towards higher-level abstraction of commodity services. One mistake we made at Motif was dedicating nearly 25% of our engineering resources building devops and supporting services rather than using commodity services like AWS (a trap that Robinhood avoided; https://aws.amazon.com/solutions/case-studies/robinhood/). We repeated this mistake by building more commodity services like KYC.
The first piece of the puzzle was to realize that companies leveraging high-quality commodity services rather than build in-house will accelerate their product development, reduce their development cost, and produce a better product. It is now possible to massively reduce the activation cost for building FinTech applications.
This Is a Recurring Problem
When I was brought in by Sequoia Capital to run product development for a Series A portfolio company, I was shocked to learn that all the processing was on a Mac Mini sitting on a fridge. As the founder dreamt of scenarios to grow at all costs, my team fought a losing battle to keep the service from crashing several times a week, catastrophic data integrity and security issues, and declining NPS scores. As I came into various companies to take over product development, the same problem popped up: the product was appallingly bad, everything took way too long to build, and the cash burn grew at an alarming rate.
What I’ve learned is that companies invest precious and expensive product development resources on commodity services that don’t differentiate the company or meaningfully add to their core value proposition. Worst yet, these components are haphazardly built with engineers who often lack domain expertise and aren’t familiar with operational concepts like person-level auditing and compliance requirements. The end result is most of product development resources are wasted on undifferentiated work that is done poorly, and the gaps are filled in by ballooning ops teams with manual labor that is error-prone and costly to maintain.
Welcome to Integration / Workflow Hell
The most sensible thing to do is to find different vendors and Frankenstein a solution together. You start with negotiating with at least three KYC vendors. Then you figure out how to do bank account verification and you’re left with two major choices: Yodlee with a more complete solution, but has suboptimal API’s and documentation vs Plaid with a much friendlier API and documentation, but woefully inadequate number of partners and dataset. If the application is a lending platform, you’re stuck navigating a maze of credit providers and going through their protracted sales and security process. If the application moves money, it’s more work trying to navigate different payment and ACH solutions, all with their own quirks.
After ⅓ of the year spending navigating and negotiating with vendors, the development team can finally start to build the integrations. Great engineers are hard to find and most of the best lack the domain expertise or operational experience needed to build a FinTech platform. Teams are stuck in workflow hell of trying to build an operational platform to monitor transactions, provide visibility for auditing and answering questions (customers tend to ask questions when they’re confused about what’s happening with their money), and tools for handling exception and reconciling data from disparate data sources.
After six months, the product is far behind schedule, the costs are ballooning, the development team is frustrated, and very little work has gone into building the company’s core IP.
The Solution is to Focus on Your Core Value Propositions
No one would define Motif Investing’s success based on our KYC feature or DevOps infrastructure, but this was something we invested significant resources in. Robinhood, Lending Club, Betterment, and other similar FinTechs didn’t succeed because they had a great money transfer mechanism. Akin to DevOps services like replication, encryption at rest, auto-scaling/auto-healing, FinTech services like KYC, money transfer, account verification, user-level auditing, ledgering systems, financial calculators, and operational tooling are all things that need to work, but none of it needs to be built internally.
AWS is the de facto standard for plugging application logic. Firebase is the de facto standard for building mobile applications. We started Productfy after nearly three decades of combined experience learning about what works and doesn’t work for building financial applications. Our goal is to provide a unified platform that reduces the activation cost for developing FinTech applications and results in enterprise-grade quality from day one.