The real spiral started here. Conversion costs had risen. Float was consuming capital. And then something broke in the middle. Our payment processing costs went up.
At $1M, you can ignore profitability. You're proving product market fit. At $10M, you can't. The math has to work. Most founders don't stress-test the math until it's too late.
A better approach: choose a much higher margin business model. Either focus on a specific vertical where you can charge much higher fees, or don't touch payments at all.
By the time you see it on a spreadsheet at 11 PM, it's usually too late. You're just deciding whether you want to die fast or slow.
The company started acquiring customers directly through ads. Not Facebook ads. That was too expensive. Industry events. Sponsorships. Email outreach. All the things that don't scale. Salespeople got hired.
Three different processors were being used as redundancy. All charged usage-based fees. At $1M monthly, processor fees were maybe $8,000. Negligible against margins.
I'd lock in pricing with customers but increase to market rate automatically. Grandfather pricing is a trap. It feels good when you're small. It's a noose when you're large.
I'd be very careful about partner distribution. Partnership economics need to make sense at scale. If the math doesn't work at 5x volume, don't do it at 1x volume.
I'd build unit economics models for every single service offering and review them quarterly. Literally every three months, ask "If we scaled this business 10x, would we make money?"
At $10M monthly, processor fees were $80,000. Still manageable, except now we were running more transactions, hitting more edge cases, requiring more support and engineering time to keep everything running.
The failure wasn't that the idea was bad. Thousands of companies need payment processing. The failure was that I optimized for growth without understanding the constraints that come with it.
Monthly revenue equals 20 million dollars
Our take (2.9 percent). 580K dollars gross
Partner fees (15 percent of our take). 87K dollars
Processor costs. 160K dollars
Payment staff (fraud, support, ops). 120K dollars
Engineering. 100K dollars
Sales and marketing. 150K dollars
I stared at the spreadsheet for a long time. And I realized we'd built something that became less profitable as it grew. The unit economics that worked at $1M actively deteriorated at $10M.
At $10M monthly? The volume of customers requesting faster payouts has grown proportionally. You now need $500K in working capital just to handle the float velocity at the level of service your customers expect.
To acquire merchants, we had partnerships. We'd work with aggregators who'd recommend us to their customers. They'd make 10-20% of the fees. Sounds fine. You're paying away some margin but you're not spending on sales.
Let's say you process $1M in payments from customer wallets to merchant bank accounts. You take 2.9% plus $0.30. Your take is $29,300. But you just moved $1M between accounts. That $1M has to exist somewhere.
We had conversations about raising prices. But we were in a market where Stripe existed. Stripe charges 2.9% + $0.30. We were charging roughly the same. Raising above that meant losing customers. Lowering meant going out of business.
At $10M monthly, the float is now $300K. And you're holding it constantly. Transactions come in. You have to pay out. There's always lag. Now you need that $300K just to operate. It's not earning anything. It's dead money.
At $10M monthly, the partners want their piece of every transaction. But the competitive market had changed. New processors had entered the market. We couldn't raise partner fees without losing partners. We couldn't lower partner fees without losing distribution.
We were processing payments for SaaS companies. Take a percentage. Keep it simple. We'd built a processing network. We'd optimized our acquisition costs down to almost nothing through partnerships. We had product-market fit. At $1M monthly revenue, life was good.
In traditional banking, the customer's bank debits their account. You credit the merchant's bank account. The two banks settle the difference through ACH. It takes 1-3 business days. During those 1-3 days, if you're the middleman moving money, you're holding the float.
This is the moment most payment startups hit it. September 2024. Staring at a spreadsheet at 11 PM. The realization. "We're dead."
I'd separate the float problem from the processing problem. Hold as little float as possible. Use the processor's settlement directly. Pay a fee for that privilege if needed. Avoid the trap of building a "faster settlement service" - that's just lending money, not processing payments.
Some customers want faster payouts. They'll pay you for next-day settlement instead of 2-day. They pay extra. So you lower the float. You move the money faster. Which means you need more capital to handle the velocity. Because faster movement means you're cycling through more float per day.
That's the death spiral nobody talks about before you're in it. It's not that you run out of money suddenly. It's that the unit economics get worse and worse until you're a billion-dollar revenue company with negative margins. And there's no escape from that without radically changing the business model.
At $10M monthly, to grow to $15M, we needed maybe 500 new merchants. Each merchant generates how much value? Let's say $3,000 per year on average in gross profit. 500 merchants is $1.5M in new gross profit per year. But to acquire those 500 merchants, I'd spent... $2M. On people. On events. On failed experiments.
So we stayed flat. 15% of all revenue going to partners. But now we had a new problem. Scale had slowed our own growth. We'd relied on partners for distribution. But partners have their own optimization problems. They might have fifty different payment options they could recommend. Why would they push ours hardest? We were just one option.
But the real problem was I'd locked in pricing with some early customers. They were paying lower rates because they'd signed up early. Now they represented 30% of our volume. I couldn't raise rates on them without losing them. And they were growing faster than new customers. So the more successful they were, the lower our blended margin became.