• Fintech Brainfood
  • Posts
  • 🧠 How is it possible Synapse and Banks can't find a $85m shortfall?

🧠 How is it possible Synapse and Banks can't find a $85m shortfall?

This issue is not a Banking-as-a-Service thing; it's a finance thing. What we see with Synapse, Evolve and Yotta is not the only example. But it's by far the most challenging to unpick.

Welcome to Fintech Brainfood, the weekly deep dive into Fintech news, events, and analysis. Join over 40,000 Fintech operators by clicking below 👇

Hey Fintech Nerds 👋

Robinhood just acquired Bitstamp for $200m. It is still aggressive in its expansion, long crypto and may have got a great deal here. 👀

This week a bank misplaced $100m, and it wasn’t Evolve or Synapse related. Reconciling accounts has always been hard. Why? That’s your Rant this week.

Visa DPS partnering with LoanPro is interesting. Will they combine this with the Pismo acquisition so you can launch an entire credit card portfolio with just one supplier?

Regulatory Nerds, you’re going to want to check out the Tweet of the Week for a knowledge drop on the first part of the 1033 rule that just dropped.

Here's this week's Brainfood in summary

📣 Rant: Why is tracking customer funds and reconciling them so hard?

💸 4 Fintech Companies:

  1. Pinch AI - The return abuse AI

  2. Equitylist - Cap Table management for global businesses

  3. Slingshot - The financial platform for creators

  4. Flex One - A modern account and treasury management platform

👀 Things to Know:

Weekly Rant 📣

Why is it so hard to keep track of $85m?

In case you missed it. Synapse, Evolve, and many Fintech customers can’t agree on where customer funds are stored. It’s now mainstream news.

In fact Synapse’s End Users Owed $265 Million — But Banks Only Hold $180 Million.

Where’s $85m dollars gone?

This week, another bank misplaced more than $100m, which had nothing to do with Banking as a Service or Fintech.

The issue is the same.

With Evolve and Synapse, the parties fundamentally disagree about the account balances. Comercia is alledged to have failed to record customer balances properly after a tech upgrade.

In both cases, the ledgers don’t reconcile

Ledgers that don’t reconcile are an issue as old as time. Banking-as-a-service makes things complex, but tracking customer funds has always been much more complex than you'd imagine.

Understanding why ledgers don’t reconcile is step #1 to unpacking what went wrong with Synapse, Yotta, and Evolve.

The Comerica story is an example of record keeping gone wrong.

A bank in Texas misplaced customer assets after a technology upgrade.

The transaction errors have forced the bank to write down any assets they can't find to make customers whole.

Comerica allows customers to withdraw funds from a trust held at UBS and instantly advances those funds to customers.

This flow of funds means customers have their cash before the bank does.

The bank has to figure that all out and reconcile it with its general ledger.

During the tech upgrade, some of these balances were not properly updated in the new ledger, resulting in their disappearance. As far as the bank knows, if the system of record hasn’t recorded a balance update in an account, that money never landed. Now they have to go back through all their payment systems and figure out what went wrong.

Every payment must match every balance, regardless of delays, timing issues, or errors.

What's this got to do with BaaS?

There's an idea that Fintech or BaaS is inherently bad. It's not. It's just complex, and it has a high customer impact.

The original sin of banking is messing up record keeping.

Recording assets and liabilities is job #1. Everyone assumes it's easy, but it's way harder than you'd ever imagine.

Synapse and Evolve is a story of records that disagree.

Evolve has a ledger, Synapse has a ledger, and the Fintech companies have ledgers. However, none of them appear to reconcile, per Jason Mikula.

Yotta, which appears to be the fintech with the largest impact to its users, said Synapse’s calculations show Yotta end users have about $111 million at Evolve, but Evolve says it’s only holding about $80 million. “The fact is that this is a house on fire,” Yotta’s attorney told the court.

Fundamentally the intermediary (Synapse) became the golden source of customer assets across multiple banks.

But the reason we don’t know who’s wrong or right is because any party could be wrong.

Because reconciling ledgers is hard.

Why? We need to look at “core banking” and general ledgers.

This helps explain the Comerica story and sheds light on why the Evolve/Synapse story is so complex.

The “core” in core banking is where all customer funds really live.

In the middle of that is a ledger. Those ledgers are supposed to work and be rock-solid. But what's changed in Fintech is that there are now many more ledgers. That creates complex funds flows that make reconciliation difficult.

Now consider that developers and engineers new to finance often assume a ledger is a database. That’s a dangerous assumption. On one level, it is, but it’s infuriatingly complex.

However, as a history lesson, understanding core banking is essential for understanding how funds are recorded. It also sheds light on both the Comerica and Synapse examples.

The “core” in core banking manages ledgers and reconciles them against payments.

"Core" and "Core banking" mean different things to different people. If you're confused, it's not you; it’s the bankers. Every banker or provider means something slightly different when they say "core."

FIS, Jack Henry, Fiserv, Thought Machine, Mambu, Tuum, and hundreds of other companies sell “core banking.” They all do different things, but if I were to generalize, here’s how it works in my head.

Building out in concentric circles from the actual core.

The Core of the Core

The core, as a system of record, Performs three main functions:

  1. A general ledger that records payments against balances. Whenever money moves, every payment has to match an updated balance in an account. This process is called reconciliation.

  2. Treasury management helps ensure that all accounts are properly funded and fully reconciled at the end of the day with central banks and third parties (e.g., other banks).

  3. Regulatory reporting summarizes all this information in a format the regulators understand and accept.

Reconciliation is much more complex than you’d imagine because institutions often update a balance before a payment arrives. This is how Chime gets paid early, or Square does Merchant Cash advance.

Core banking providers do much more than these three things, but this is what sits at the core of the core. Notice that I didn't mention payments, yet some people consider their payment processor to be their "core."

That's because many payment processors are also core banking providers (like FIS, Jack Henry et al.).

Confusingly. Some payment processors help maintain a system of record for card issuers (e.g., Galileo, Marqeta, Q2, etc.).

Each of these companies has a patchwork of products but anchors in a different part of the stack.

Payments and Money Movement

At an abstract level, money moves in one of three ways.

  1. Domestic: local payment methods in the US, like ACH, Fedwire, or closed wallets like Zelle, Venmo, and PayPal, all fall into this category. Globally, there are hundreds of equivalents, each with its own unique quirks and idiosyncracies.

  2. International: Historically, this meant SWIFT (or, to US readers, "international wires." Increasingly, Visa B2B Connect and other services like PPRO help companies connect directly to another country's domestic payment systems (e.g., help a Fintech in the US connect to Faster Payments in the UK, UPI in India or Pix in Brazil).

  3. "Cards" is the general name given to debit and credit card "pull" payments. What makes these unique is the payment instrument (AKA the card).

Core banking platforms often support each of these payment types, either directly or as a "module." Banks and nonbanks may also choose to get the capabilities from a specialist provider or build in-house.

One bank might manage checking accounts, core banking, and domestic, international, and card payments through FIS or FiServ.

Another might have an in-house solution for core banking, domestic, and cards (e.g., Nubank and Monzo).

That's why I don't include payments in the core of the core. Because, in reality, payments sit outside it.

Remember, each of these "payment systems" is a messaging system. Money only moves in a legal sense when the general ledgers of all the banks in a transaction have fully updated and reconciled their customer accounts.

Value Added Services

Smaller banks often rely on (and sometimes feel prisoner to) their core provider for everything. This can include:

  • Channels: e.g., Online and mobile banking

  • Fraud prevention: Across cards, ACH, customer onboarding

  • Customer onboarding: Including checking credit history, creating a customer account, and being the "CRM."

  • Analytics and marketing: Customer segmentation, email creation, basically "Hubspot + Big Query + Tableu for enterprise with a terrible UI."

  • Statements and card production: The companies that print paper and physical plastic need to integrate with them. The core providers often remove this effort from the shoulders of their smaller banks.

  • etc

For a "core provider," the value-added services can account for 30 to 40% of overall revenues.

If you're a Fintech reader, a critical thing to keep in mind is these platforms are usually "one-size-only." If you want any change or to use the product for a unique use case, generally, that will be a project you have to pay the provider to do.

The core provider becomes the technology team for banks.

Fintech has unbundled the core provider to some extent. Where the D2C or D2C Fintech company manages the value-added services, the BaaS provider manages payments, and in theory, the bank still manages the ledger. But also, now everyone else has a ledger, too.

Complexity: The Product Segment and Geo

In reality, checking, loans, and payments look wildly different for consumers, businesses, and wealth management. There aren't many challengers (younger) "core banking" providers capable of handling the needs of complex global corporations (yet).

That's why the inside of every giant global bank has 1,000s of systems. Many of them are built to solve

  1. For a product type (e.g., loans, checking, payments)

  2. In a given market (e.g., US, UK, India)

  3. For a specific segment (e.g. Corporate, wealth, consumer)

Now consider how many banks are made up of M&A of many smaller banks. Being a "core provider" is not simple.

Why do banks use legacy cores?

The reality is that the core providers sell one thing, above all else: stability.

a) Because the consequences of failure are enormous. British bank TSB was fined £49m after its core banking upgrade failed, preventing millions of customers from accessing funds or paying bills. BB&T scrapped a core banking upgrade entirely.

Customers unable to access accounts or pay bills is the kind of catastrophic failure that results in mainstream news coverage, regulatory fines, and, in some cases, CEO job losses.

b) Stability in the general ledger is job #1. On the surface, a ledger performs a simple job. It should reconcile payments against accounts. Every customer (and the bank) has an account, and those balances get updated when a payment either increases or decreases the balance.

Simple right?


Take this "simple" card payment.

The "Flow of funds" complicates everything. Even a simple payment, such as the card payment above, can be complex, as it requires recording and reconciling at the merchant, the card network, and both banks. To complicate things, the gap between the payment and when money moves is often days.

When you layer on interest rate calculations, foreign exchange, and dealing with settlement delays, complexity increases exponentially. (Mapping flow of funds and edge cases could be an entire book).

c) Stability in regulatory reporting. The system of record represents the "truth" for the regulator. The output from the core is often sent directly to regulators (or with light editing).

There are hundreds of laws financial services companies are subject to that vary by jurisdiction, product, and customer segment. Often, a legacy core provider will have done this before for another financial institution (likely several).

It's this quirk that allows some people to believe that a given core provider is "regulated."

There's no such thing as regulated software, but there is software that regulators feel more comfortable with.

I’m not saying legacy cores are the only answer.

I’m saying getting reconciliation right in ledgering is non-trivial and deserves a lot more attention.

Reconciling one ledger is hard. BaaS makes it a multi-ledger problem.

The Comerica Bank example from the introduction is a classic example of what can go wrong when a bank tries to replace one system of record for another. A payment system "rip and replace" can be catastrophic.

What we saw with Synapse is that it appears to Evolve, and the Fintech companies relied on Synapse as their general ledger. The BaaS provider was helping each Fintech company see accounts at multiple banks. Meaning the BaaS provider (or Fintech company) was the only one who had a 360 view of the customer.

This is common.

Often, Fintech companies either build their own “core” or rely on someone else’s (like the BaaS provider). This creates a very complex flow of funds (as Omi from Fragment.dev helpfully illustrated in the image below).

Every single box in this picture has a ledger. Every single one of them should reconcile. But if the bank and the Fintech company rely on the “BaaS” provider, and

When the ledgers don’t reconcile, things go wrong.

The Synapse ledger doesn’t match the ones at Evolve or the Fintech companies.

This has created the need for forensic accounting to determine where the balances are or were.

Reconciling a ledger is hard.

But it’s a job #1.

Now, we're seeing banks opt to launch new products on a new core, either for direct new product launches or, in many cases, to enter Banking as a Service or Embedded Finance.

This solves the massive risks presented by rip and replace, but it presents new risks.

We need better Ledgers and Reconciliation to fix BaaS and Embedded Finance.

Issues reconciling the ledger.

That's an issue as old as time.

Having Fintech wallets that don’t agree with bank ledgers is a new issue that needs a thoughtful approach from providers, Fintech companies, and regulators.

Understanding how money moves and what makes it complicated is critical before we can unpack how that applies to BaaS.

Most Fintech companies will use more than one bank as their partner.

If they want to offer debit cards, credit cards, high-yield savings, and stock trading, they might need 5 different banks or brokerages.

Now, you have a situation where no single bank or brokerage has the full customer picture. The only people who would be the intermediary (e.g., the BaaS provider) or the Fintech wallet.

This fundamentally means it doesn't matter how new or old the bank's core systems are; nobody has the full picture except the Fintech company.

Between the CFPB's approach to wallets and the federal agencies’ actions on third parties, the answer appears to be to regulate wallets and banks more. Yet we don’t know when the CFPB’s wallet rules will come into full force, and in the interim, at least, we have no clarity on how these issues can be solved if they happen again.

The only possible solution is for Fintech companies to get great at ledgering.

Fortunately, plenty of Fintech nerds love a flow of funds diagram and new core banking providers specializing in BaaS.

There’s also ledger-as-a-service and reconciliation specialist Fintech companies.

Wherever there is pain, there’s opportunity.

But my biggest takeaway is that I hope we’re learning lessons on the importance of getting the basics right.

Balancing accounts, reconciling a ledger against payments, that’s job #1.

If you want to be great at BaaS, get great at ledgering, or find someone who is.


4 Fintech Companies 💸

1. Pinch AI - The return abuse AI

Pinch AI helps e-commerce merchants optimize for checkout conversion but spot customers who could be potential "returns abusers." The service monitors the user's devices and behavior patterns and comes with pre-loaded rules to spot possible return abusers.

🧠 Returns abuse has skyrocketed since the pandemic. From simple wardrobing (buying clothes, wearing them once, then returning them as "faulty") to more complex, like empty boxing (where a fraudster returns a box of the right weight without the item). These attacks are hard to spot for most merchants who have never seen their customers before. This is a huge problem, so it makes a great wedge product for Pinch, but can they scale it out from there?

2. Equitylist - Cap Table management for global businesses.

Equitylist lets companies issue and track stock, manage complex cross-border shareholdings, visualize dilution, and manage equity compensation. It also provides 409A valuations for companies based in India, Singapore, and the US.

🧠 US-based VCs are the world's VCs. During the pandemic, America exported venture capital, and one primary beneficiary was India. India has seen an explosion of entrepreneurial activity in the past half-decade, building on a historically digitally savvy and well-educated population. Now, they can serve the fastest-growing home market and operate remote businesses. This is a spin-out from Angelist India, so it is not a "startup," but it is still an interesting use of Fintech.

3. Slingshot - The financial platform for creators.

Slingshot helps creators run their small businesses, withhold a % of income for tax, and access healthcare and retirement funds. It also combines bookkeeping, centralizes payment data, and links it all to the primary card. Under the hood, they use Stripe Issuing and Stripe Treasury. (h/t Techcrunch article)

🧠 There have been a few "creator Neobanks," most famously Karat. We don't see these names in headlines often, but I've heard a few people say, "Karat is doing quite well." (Whatever that means). There's certainly no lack of creators as a profession, and being a freelancer has always sucked. This is a space that is not well solved because few have made it profitable. Maybe that time is now?

4. Flex One - A modern account and treasury management platform

Flex provides deposits, checking, credit card, and "banking." The service provides dashboards to allow companies to sweep cash between accounts and maximize their available cash on hand vs yield. The service has a flat 1% currency fee on all international wires, making it tempting for global businesses.

🧠 There's a slew of these companies. Flex is a little bit like Rho, Mercury, or Arc. I recall when Rho first arrived they differentiated towards FX and cross-border but seem to have dialled that back. This space is so crowded I made a whole podcast series about it. Flex has a premium feel to its branding and is leaning on "banking experience." Perhaps there's a little more white-glove here than first appears?

Things to know 👀

1. Monzo just released its FY2024 results. Spoiler alert. They crushed. 

Monzo delivered

  • - £15.4m ($19.16m) profit

  • - 2.5x revenue growth

  • - 88% increase in deposits (to £11.4bn or $14.51bn)

  • - 47% increase in card spend (to 47.8bn or $60.85bn)

  • - 84% increase in the gross loan book

🧠 Three things have driven the Growth.

  1. An increase is in customers to 9.3m consumers (from 7.4m) and businesses to 400,000 (from 200k a year ago).

  2. New product launches like tax-free savings (ISA) now have 1.3m customers benefitting from higher rates, use of overdrafts and "Monzo Flex" their take on a credit card/BNPL. They have 500k customers paying for their subscription product.

  3. More profitable customer cohorts as customers get older, their average deposit balance increases. An "elderly millennial" is now 42 and probably makes a lot more than they did when Monzo was founded in 2015. Businesses are more profitable than consumers, etc.

🧠 Growth will also come from three places.

  1. Customer acquisition. They expect to hit 11m consumers by this time next year.

  2. Product extension. Monzo hasn't started towards mortgages or deeper forms of business lending. They have a ton of headroom here.

  3. Market entry. Monzo plans to enter Europe via an Irish HQ in the next 12 months and has a team in the US looking to open that market.

🧠 I think they'll win on UK customer acquisition and product extension, but the jury is out on global expansion.

🧠 Can they expand to Europe? Europe has a patchwork of players like Bunq with some geographic presence, but my question is, who's Monzo's first best customer outside the UK, and why? It may be the Republic of Ireland for precisely the reasons it worked in the UK, but then where next?

🧠 Can they expand into the US? They're committed to doing so, but it is such a competitive market. The bank license (charter) and owning the infrastructure (vertical integration) have helped them win in the UK, but that's much harder to do in the US. I'd be tempted to start with something more business or freelancer-focused (like their UK SMB proposition).

🧠 I don't see many domestic threats. Their direct competitors, like Starling, are also growing and having a fantastic time, and the incumbents are too slow to react. Most incumbents are not growing revenue faster than inflation, and their share price growth is driven by the higher rate environment, dividends, and share buybacks.

Loved this nugget: Customers have an average of 38 friends who also use Monzo.

Robinhood says the goal is to help it expand internationally and win institutional business. Based on its last accounts Bitstamp was doing roughly $30m of revenue and was profitable.

🧠 Will it work? Robinhood did at least a solid job integrating X1 so I wouldn't bet against them. Kraken is strong in Europe, but this gives Robinhood a foothold in Crypto here. The Bitstamp product is quite simple for consumers.

🧠 Robinhood got a decent deal here. At 4 or 5x revenue, that's not a bad price. Consider Bitstamp had ~4m users. That’s close to $50 per user acquired (although it’s not clear how many of those were active).

🧠 Robinhood isn't backing away from crypto it's charging towards it. This consistency could be huge for them in time. They're still in expansion mode, and look like much more than the day trading app they were at IPO. You’re probably still not paying enough attention to crypto.

Good Reads 📚

Alex Johnson had a great quote: "You never would have imagined helping companies manage their finances would become the hottest space in Fintech, but it's what makes this industry special." People are attacking it from different angles with Navan coining the phrase "BYOC" or Bring your Own Card. Ramp just launched "Ramp travel."

🧠 Everyone is doing a bit of everything in the race to be the software operating system for growth companies. Spend is the perfect wedge.

🧠 The other route is treasury management first, spending second (Like Mercury or Arc).

🧠 None of these are wrong and that's what makes it fascinating. This is exactly why I wanted to interview the leaders of this category.

You can find the first three episodes of Fintech Brainfood Interviews on Apple Podcasts, Spotify, or wherever you get your podcasts now.

Tweets of the week 🕊

That's all, folks. 👋

Remember, if you're enjoying this content, please do tell all your fintech friends to check it out and hit the subscribe button :)

(1) All content and views expressed here are the authors' personal opinions and do not reflect the views of any of their employers or employees.

(2) All companies or assets mentioned by the author in which the author has a personal and/or financial interest are denoted with a *. None of the above constitutes investment advice, and you should seek independent advice before making any investment decisions.

(3) Any companies mentioned are top of mind and used for illustrative purposes only.

(4) A team of researchers has not rigorously fact-checked this. Please don't take it as gospel—strong opinions weakly held

(5) Citations may be missing, and I've done my best to cite, but I will always aim to update and correct the live version where possible. If I cited you and got the referencing wrong, please reach out