A few weeks ago, I shared a thread on Twitter that went semi-viral (2.4k likes!) It was called ‘some lessons from building a personal finance startup’ and it was about my experience spending 5.5 years building a personal finance app – Birch – that was ultimately acqui-hired.
Before this post, I always received lots of requests from consumer fintech founders asking for feedback, but it’s gotten even crazier since this tweet made the rounds.
I don’t have the time to meet with every single person and give feedback, so I’m sharing my tweetstorm here to use as a resource – most of my answers to common questions are in this thread:
Some lessons from building a personal finance startup:
— Alex Cohen (@anothercohen) July 22, 2020
Building a personal finance app is a lot harder than most people realize. Lots of handwaving on Twitter about building the next Mint, but even with Plaid/Yodlee/etc you have lots of things to “fix” before you can even launch
— Alex Cohen (@anothercohen) July 22, 2020
Things such as normalizing merchants (making sure Uber1237ahd72 returns as Uber and UberEaaaats092220 returns as UberEats) and categorizing (is Uber ridesharing or taxi or travel or ???) are hard and have to be solved before launch
— Alex Cohen (@anothercohen) July 22, 2020
Then you deal with the issue of different FIs returning different lengths of transaction histories. Chase used to return only 3 months, as an example, while Amex would *sometimes* give us 2 years of history, depending on the account ¯\_(ツ)_/¯
— Alex Cohen (@anothercohen) July 22, 2020
So when you’re trying to analyze someone’s transaction history to make recommendations, you’ve got a whole clusterfuck of inconsistent data across providers. So what do you? Take the shortest historical range across all accounts and project out a full year? Or ??
— Alex Cohen (@anothercohen) July 22, 2020
Bank connections also break constantly. Be prepared to deal with endless customer support tickets asking why new transactions aren’t showing or why the app has asked them to reconnect Barclays 17 times
— Alex Cohen (@anothercohen) July 22, 2020
It’s also really, really hard to make money as a personal finance app.
— Alex Cohen (@anothercohen) July 22, 2020
You have a few options:
– Charge a monthly fee to your users
– Push for affiliate revenue
– Offer your own financial product (like an installment loan or debit card)
The problem with “charging your customers ” is that it’s awkward to ask users to pay money when your entire value prop is saving them money…
— Alex Cohen (@anothercohen) July 22, 2020
Some are doing this in other ways – $10/mo for coaching or $xx/mo for additional bank connections. I’m not sure what conversion is
Every PFM then has the magic ‘aha’ moment. If we make the _best_ recommendations people will open up new accounts through us and we will get paid. Win for the customer, win for us, win for the banks, right?
— Alex Cohen (@anothercohen) July 22, 2020
No, wrong
99% of the banks you actually want to partner with – Chase, Amex, Capone – don’t care about the extra 25 or 50 accounts you’re going to drive to them per month when you’re just getting started.
— Alex Cohen (@anothercohen) July 22, 2020
So you have… 2 options
Either 1) partner with shitty banks with bad products because they will work with you or 2) make an amazing recommendation and then send your user to another affiliate site to have to find and click that product _again_ – a bad experience
— Alex Cohen (@anothercohen) July 22, 2020
3) is probably your best bet if you have the capital and partnerships to actually launch a product, like a debit card or savings account. Your PFM becomes user acquisition for your product. But very few startups end up being able to do this and this comes with its own challenges
— Alex Cohen (@anothercohen) July 22, 2020
The other thing about building *anything* in the PFM space is that it becomes really fucking expensive to acquire customers. Everyone is fighting over the same ad real estate.
— Alex Cohen (@anothercohen) July 22, 2020
You’ll likely spend upwards of $15-$30 to acquire a customer and get them through your funnel
Speaking of acquisition, enjoy a nice subset of your users dropping off during the signup flow because they can’t remember their bank logins and passwords.
— Alex Cohen (@anothercohen) July 22, 2020
No one remembers their logins and passwords
The last thing I’ll say is if you are launching a PFM, expect your costs to be much higher than you anticipate. Plaid and Yodlee are fantastic businesses and their APIs are expensive.
— Alex Cohen (@anothercohen) July 22, 2020
I guarantee you’ll consider building versus buying at some point early on
Between the cost of building your apps, API access, and the difficulty of making money, you better have some really innovating acquisition model or monetization model or you’re going to have a bad time.
— Alex Cohen (@anothercohen) July 22, 2020
Between the cost of building your apps, API access, and the difficulty of making money, you better have some really innovating acquisition model or monetization model or you’re going to have a bad time.
— Alex Cohen (@anothercohen) July 22, 2020
Good luck!
