Hi everyone,
I’ve worked in fintech for the past 12 years, and for the last 4 of them as an early-stage fintech investor. I’m still learning every day as GP of The Fintech Fund, and – excitingly – the fintech space continues to grow and evolve in surprising ways.
One of my goals is to share my own learnings along the way for the benefit of TWIF readers. I’m sure that not all of these fintech insights will be novel or groundbreaking, but they’ve been meaningful to me, and I hope you’ll find them interesting.
This will be an irregular series to break down thoughts and lessons from my experience as an early-stage fintech investor and operator, as well as reflections for founders and other investors. As always, I welcome any pushback and feedback!
Today’s brief reflection is on revenue scalability.
Companies covered here include SaaS fintechs like Themis, Easy Expense, Opareta, GuruHotel, and Payable, API aggregators like Stitch, TrueBiz, Zenpli and Fragment, transaction facilitators like Paytrix, Ansa, and Rainforest, banking providers like Griffin, Octo, Vault, Archa, and OutGo, and lenders like Ampla, Yave, and Flychain.
One of my least popular opinions with the fintech crowd – and a controversial one for a fintech investor – is that ‘fintech’ is not a real product category.
The term fintech describes a horizontal collection of somewhat-related products, covering anything from e-commerce logistics, to proof of identity tools, to subscription payments, to core banking software, to consumer microloans in emerging markets.
It’s almost impossible to build a comprehensive fintech market map (though I’d love to see someone try). CB Insights publishes their wide-reaching Fintech 250 every year, and even that leaves out entire swathes of the industry, by necessity:
The reason I bring this up is because not all fintech business models are comparable.
Something I focus on when investing is the scalability of revenue. The question I like to ask is, “once the product is built, how easy, cheap, and lucrative is it to acquire a new customer?” How much does it cost to earn $1 more?
This is, of course, an old and foundational concept for SaaS investors, where the “rule of 40” convention dictates that high-performing companies’ annual growth rates and profit margins should add up to 40%. ie: Thin profit margin of 2%? Then your company should be growing at least 38% YoY.
This heuristic is helpful, but the reality is much more nuanced. And it’s particularly relevant in fintech, because so many companies in fintech are not SaaS companies or anything close to it.
Fintech has a broad diversity of revenue streams, which I generally like to rank in order of scalability cost.
These are the five most frequent revenue streams I see from fintech companies, ranked in order of scalability cost:
Lending
Banking
Transactions
API calls
Software
This isn’t a complete list. A notable omission is advertising / referral commissions, the primary revenue type enjoyed by Credit Karma, Nerdwallet, MoneyLion, and others – I’ll come back to these types of businesses in a future writeup.
(I am bad at graphic design)
1. Lending
Why does lending revenue tend to be expensive to scale? Many reasons.
Other than the generic costs that every company incurs to acquire users (CAC, onboarding costs, COGS, etc.), lending revenue has specific features that make it expensive to scale:
Most fintech companies lend from debt warehouse facilities they raise, which carry a fixed rate they need to pay back (eg: 12%), so for every $100 in loans made, the fintech company needs to pay its facility provider $12 before it earns $1. And this is usually true at the portfolio level, not the loan level: if a fintech company makes two $100 loans and one borrower only pays back $100, it is required to pay $24 back to the facility provider before it collects $1 in revenue. (This is a very stylized example - there are nuances.)
Many fintech companies experience at least a nominal amount of fraudulent applications, which eat away at loan margins.
Lending accounts tend to carry fixed servicing costs to manage, either from an internal customer service team or a third-party loan service provider or processor.
Fintechs are normally required to contribute equity to their loans – especially early on – in order to have skin in the game. ie: For every $10 lent, the debt facility contribution may be capped at $8, while $2 has to come off the fintech’s balance sheet. This is capital that could be going towards hiring more engineers, scaling marketing campaigns, or building new products - so it is an expensive use of capital.
There are other variable costs of making and collecting loans: card printing and replacement costs for credit cards, regulatory and licensing costs, statement printing costs, etc.
This is not to say that lending revenue is “bad” revenue - at scale, a good lending business can be very lucrative! But each dollar of revenue comes with significant costs, which tend to drop at scale as a % of revenue, but which initially are very expensive.
Our lending portfolio companies at The Fintech Fund include companies like Ampla (banking and billpay for consumer brands), Yave (mortgages in Mexico), and Flychain (factoring for home health businesses).
2. Banking
Now consider banking: what are the costs to acquire new deposits? In most cases, less than lending, but there are still fixed account costs, KYC costs, regulatory costs, fraud costs, and servicing costs for each additional bank account.
Costs tend to be material and scale linearly with the number of bank accounts (though the value of each account may outpace costs in terms deposit growth – ie. a marginal customer may bring anywhere from $100 to $100,000 in new deposits).
Again – this does not mean banking revenue is bad revenue: most banks are making a killing this year on collecting Net Interest Margin alone! But it is not costless revenue to scale.
Our banking portcos include companies like Griffin (banking-as-a-service for the UK), Octo (banking for Egyptian consumers), Vault (corporate banking for Canada), Archa (corporate banking for Australia), and OutGo (banking for freight carriers).
3. Transactions
Transaction-based products are in the business of monetizing massive quantities of transactions by charging per-txn fees. Examples can include anything from credit and debit cards, digital wallets like PayPal and Venmo, stablecoin payment platforms, cross-border rails and payment orchestrators like Nium and CurrencyCloud, invoicing products like Bill.com and Quickbooks, etc. Transactions tend to be low-margin, but the high volume and instant revenue recognition make them an attractive source of revenue.
The MaRS Discovery District does a good job breaking down the differences between recurring revenues (which software products tend to enjoy) and transactional revenue:
Transactional revenue models are based on predictable sales of goods. Transactional revenue models are less attractive than recurring revenue because a company has to “do” something anew for every sale (produce and ship goods). Toothpaste and printer toner provide good examples of transactional-revenue products.
Instant revenue recognition is a very attractive feature of transactions: if you process a credit card transaction for a customer, you get paid instantly. Meaning that as soon as you incur the ‘cost’ of making the product (ie: paying out network and interchange fees to your financial institution partners), you also recognize the revenue of ‘selling’ the transaction. This is preferable from an accounting and forecasting perspective to selling software and then waiting for monthly or annual bill payments, which it is always possible may be challenged or unpaid.
With that said, transactions also tend to have costs that scale linearly with revenue. Your transaction costs may be a % of revenue, meaning for every $10 in revenue you’ll always pay (eg) $5 in txn fees, or they may be a fixed set dollar amount per transaction, such as ACH fees, which incentivizes fintechs to push their customers towards larger transaction sizes in order to increase their take rates.
Transactions tend to be fairly low-margin compared to SaaS products (think: 1-2% take rate on cards vs. 40-80% gross margin on software), which creates winner-take-all dynamics in that transaction-based businesses which hit critical scale can be extremely profitable (and the scale + network effects create a moat against competitors), but the scaling itself is usually precarious and expensive (such as incentivizing new cardholders with sign-up bonuses).
Our transaction-based portcos include companies like Paytrix (payments curation for European businesses), Ansa (closed-loop payments platform and digital wallet for businesses), and Rainforest (payment processing for brands).
4. Fintech APIs
Financial products that sell APIs look similar to software in a lot of ways – customers are paying for access to a platform, which in turn gates underlying data that it makes available to the customers on either a batched basis or a per-use basis. Examples in fintech are data aggregators like Plaid or Alloy: they create connections to underlying data providers and sources, call the data on behalf of customers, clean and normalize it, and present it in a useful format for the end-user application.
Once the heavy lifting of creating the API connection to the underlying data is done, APIs are relatively costless to maintain and scale. Naturally, they will constantly break down as the underlying provider makes changes to their own product, and this requires constant QA and repair, but once you have a connection in-place, you can sell 100 or 1 million API calls and incur close to the same costs (the expense of calling an endpoint is usually nominal to the aggregator).
Again, like transactions, API aggregators are in the game of scale: because they pay high fixed (setup) but low variable costs, they price smaller tiers of API calls at a higher per-call rate, and give discounts for higher call volumes (sometimes requiring fixed monthly minimums), in order to increase their overall take rate from each API.
Our transaction-based portcos include companies like Stitch (open banking aggregation and payments initiation for Africa), TrueBiz (a know-your-business provider for fintechs), Zenpli (a know-your-customer aggregator for Latin America), and Fragment (a universal ledger and database for money).
5. Fintech SaaS
The last fintech business type that I outlined above – and to me as an investor, the most attractive – is software-as-a-service. Examples of fintech software include accounting, compliance, financial forecasting, investing, secure communications… basically any digital financial product with a recurring monthly cost. Examples include Bloomberg, Quickbooks, Aladdin, Addepar, etc.
There is an incredible wealth of information on the attractiveness of SaaS business models, but as a fintech investor, this is what makes these businesses compelling to me:
High gross margin: Normally, there is a fixed cost to build SaaS products up-front, but once they are built, they can be scaled infinitely with very low (or non-existent) variable costs. This makes them extremely lucrative and allows SaaS companies to price at gross margins normally from 30-100%, depending on how competitive the product space is.
Low scaling costs at-scale: In the hypergrowth phase of building a SaaS product, when the market is reacting positively to the product and customers are signing up in droves, these products take a lot of investment to quickly build and maintain. At scale, these cloud-based products are very inexpensive to maintain and scale almost infinitely.
Defensive positioning: Often, the feature functionality of SaaS products and IP that goes into them creates a moat that deters new entrants in the space from stealing market share from SaaS providers, creating winner-take-all dynamics for the products that scale the most quickly.
Our SaaS portcos at The Fintech Fund include companies like Themis (a compliance workflow stack for banks and fintechs), Easy Expense (simple accounting software for small businesses), Opareta (Shopify for African agent-based businesses), GuruHotel (an automated web developer for hotels), and Payable (treasury management for modern finance teams).
