We rip on the payments industry a ton. And deservedly so. But somewhere in the ranks are people providing good, honest service with tangible value. So for those of you who feel slighted, our comments aren’t meant for you.
Still, all of this begs the question: what payments margin is deserved? How much margin is too much? At what point does payments just become absurdly priced?
We need to discuss this from two perspectives. First, the persona of the payments provider, and second, with respect to a software landscape.
There are basically two types of people that sell payments: payments bros, and solution providers. The former are known by many epithets: trunk-slammers, wham-bam-thank-you-mam-ers, et al. They sell payments, offer no other value, and definitely can’t be counted on for support. Hand-to-God they spend their working moments lying to get a signature and what happens after the ink dries is someone else’s problem.
The latter have real costs because they provide real value. They may have gotten into payments processing due to business model changes, or may have realized that selling value was better than competing on short term price. One such example of a value provider who stumbled into payments is the POS reseller. The abbreviated history of the POS goes like this:
- 20 years ago the reseller could expect 40% margins on hardware, 50% margins on software (which was a one-time, non-recurring expense), and installation work (often called PIT for programming, installation, and training) was about of the revenue stream. For a system that cost $30,000 the dealer might recognize a gross margin of $21,000
- 10 years ago margin compression starts. Hardware margins fall to 25% as Asia enters the scene, software still isn’t SaaS so the margins remain at 50%, but the POS system is now asked to do more and requires talking with other systems (OpenTable, ecommerce, etc). Now PIT becomes about 25% of the total cost. Total system cost falls to $25,000 but the gross margins have fallen a bit too, down to $12,000.
- Today the hardware margins are a wash – 10% at most . Software is now monthly, meaning there’s no upfront margin, and the cloud systems require less PIT because they’re just built so much better – estimates we’ve seen peg cloud POS PIT at about 30-40% of traditional on-prem POS work. Today the price to place that $30,000 system is now less than $10,000, and of that the only margins come on the PIT, but it’s likely less than $2,000.
Payments have come to fill the revenue gap for POS resellers. And in reality, resellers were already doing most of the payments support, even if they weren’t being paid for it; with the payments stream at least resellers are being compensated for work put in. Dealers we talked with estimate that 30% of their support volume is payments-related. Often these are simple fixes from merchants doing dumb things, but sometimes there are real issues: downed networks, chargeback research, fraud, and more. For this persona the payments stream offsets some of the costs for supporting payments.
Still, the rules of thumb on payments costs look pretty lucrative, even for the POS dealer. In the POS world you need a FTE (full time employee) to support every 125 customers. Let’s say a support tech costs $70,000 per year. Across 125 sites that’s $560 in annualized support per site to cover costs. In payments, the rule of thumb is one support rep for 700-800 merchants. That’s literally a near exponential leap. Translating that to dollars means a meager $87.50 in annualized support cost per payment customer.
Let’s take a merchant who does $500,000 in annual revenues. We’re going to assume a Lightspeed acquiring margin of 40 basis points, or 0.4%. Let’s say a dealer splits this 50/50 with a sponsor ISO who’s taking their share for underwriting the risk (remember that Lightspeed is a Payfacs so they command higher payment margins than most alternative payment models).
That’s still $1,000 a year to the dealer, or $83 per month.
Using our payment support math, that basically means 11 months of payments revenue are free money since the COGS has been taken care of. Which now gets us to comparing payments against software businesses, so let’s do that.
If you were to begin a payments business you might be inclined to sell payments yourself. But if you’re a POS dealer you probably have your hands tied and need to hire a sales person instead. A payments rep is going to earn 5x the upfront payments residual as well as 20% of the ongoing residual. It sounds a bit confusing so we’ve made a table to clear it up.
Assume a rep does 10 deals a month and you have no churn. Each deal brings in $100 per month. As you can see the rep doesn’t start contributing money to the business until month 7, and even then the business needs to wait until month 11 to break even from a cash position on what it cost to hire the rep.
You should also consider that you’ll need money to buy merchants out of their early termination fees, or ETFs. ETFs shouldn’t exist, but merchants are dumb, so there might be $150 per ETF that the POS dealer must consider to win over each account. Over a year that’s maybe another outlay of $18,000. But don’t stop there.
Churn in payments is pretty low. While people will buy merchant portfolios for 30x monthly revenue, implying a 30-month life of a merchant, our data shows that merchant payments churn is only 15% a year. Let’s go ahead and carry this math out for 30 months to see what it looks like.
Over 30 months that one payments rep has contributed $262,000 to the business, signing up 300 merchants. That’s not including the salary of the sales rep made for himself, which at $7,200 per month annualizes to $86,400.
Now we should compare this to software. Fortunately we wrote a cost model for POS software previously. POS is a bit of an extreme, but it’s going to take at least a year to build any sort of respectable software product. The level of complexity may vary, but it’s at least 2 engineers working full time on the solution. Over a year that’s $300K in COGS since an engineer is $150,000 a year at a minimum.
Then you need to think about selling the product. Engineers are notoriously bad sales people. So now you need to hire a sales person. Let’s say the sales person costs $85,000 per year just to be consistent with our payments example.
Unlike payments, which everyone must take, software must deliver demonstrable value to attract customers. This means that the sales rep is not going to sell 10 deals per month, but more like half that amount. Getting $100/mo from a merchant is about the limit of what’s possible in 95% of cases (the outliers are the delivery or daily deal businesses, but those are the extremes). Meanwhile the engineers are still needing to iterate on features and squash bugs, so their salaries don’t disappear in year two. Here’s that table for comparison, excluding the $300K it took to even build the product.
We also need to think about churn. In SMB software, churn is 4% per month, or nearly half your customers per year. This is a 3x increase over payments churn. We rationalize the difference two ways:
- Merchants must accept payments and therefore have to pay someone this amount
- Payments get deducted from revenues automatically so merchants don’t even feel like they’re paying for it – i.e. they’re not getting a credit card bill for the subscription
Obviously these are drastically different financial outcomes. You’re probably looking at the numbers and wondering, “Who the hell would bother selling software to SMBs given these grim financials?”
Congratulations: you’ve just answered the reason why investors won’t back brick and mortar solutions and why merchants, correspondingly, have no software worth a damn.
Payments just become more and more accretive over time. A large ISO told us that their customer acquisition costs (CAC) were less than $100. COGS are another month of revenue, so it means you’ve paid back CAC + COGS within 2 months! SaaS businesses take 12-18 months to pay back CAC on average. SaaS businesses also shoot for a customer lifetime value (LTV) greater than 3x CAC. If a merchant has an actuarial payments life of 30 months, LTV is not 3x CAC but 30x CAC. This time it’s a literal exponential improvement over software.
So what should payments margins be?
Considering it adds no value relative to software and has drastically lower support costs it would seem that the pricing should be much less than it is. Another way to think of this is that payments is a business with 92% margins. That’s crazy. That’s even higher than the margins for alcohol production. Those providing higher value with a payments stream can’t mathematically justify the margins on a product-basis, even if it’s filling up some revenue shortfall from the conventional reseller model.
The high cost of payments has prevented merchants from using those funds to acquire software that delivers tangible value and helps them succeed. But merchants are too dumb to understand it. And we don’t see that changing for a while.
Add comment