The problem with treating BNPL as just another payment method
Adding to the checkout feels like an easy win. Affirm, Klarna, or Afterpay lets the customer split the payment. More shoppers say yes. Soon a quarter of your orders flow through it. Founders call that proof. ‘A quarter of our customers use it. Clearly it works.’
Here is what the usage number hides. BNPL charges you 5 to 6 percent of the order. That is about double the card rate. Some BNPL buyers truly needed the split payments to buy. But many others were buying anyway. They just picked the shiniest button at checkout. On those buyers, you paid double fees for a sale you already had.
Usage tells you how many people use BNPL. It tells you nothing about how many people bought because of it. And almost nobody measures the difference.
This article is about the fourth one: sales channel and payment fees. BNPL is the most expensive line in that stack. The honest test is not how many customers use it. It is how many customers it truly brings you.
1. An example showing you the numbers
Say your store sells mid-century furniture online. Your is $500. Two years ago you added BNPL. Today 5,000 of your 20,000 yearly orders go through it.
On every $100 of sales, BNPL takes about $6. A card would take about $3. On your $500 furniture order, that is a $30 BNPL fee. Across 5,000 BNPL orders, it is $150,000 a year.
Then you run the incrementality test. It asks one question. Which of those 5,000 buyers truly needed the split payments to buy? The answer comes back at about 30 percent. The other 70 percent would have paid by card. Here is what that split does to the $150,000.
The $150,000 BNPL fee bill, split by what it actually bought
Same store, same providers, same 6% fee. The test just sorts the orders.
| Line item | All BNPL orders | New buyers (incremental) | Would have paid by card |
|---|---|---|---|
| Orders per year | 5,000 | 1,500 | 3,500 |
| BNPL fee per $500 order (6%) | $30 | $30 | $30 |
| Yearly BNPL fees | $150,000 | $45,000 | $105,000 |
| What the fee bought | - | a new customer, at $30 each | nothing, the sale was yours already |
Read the two right-hand columns. The $45,000 on the extra orders bought you 1,500 buyers you would not have won. That is $30 per new customer. Cheaper than most stores pay on paid social. That money earns its keep. The $105,000 on the other 3,500 orders bought nothing at all. Those customers were already sold. BNPL just collected double fees.
One honest note. The $105,000 is a leak. It is not an instant saving. Pull BNPL off the checkout and you may scare away some of the 1,500 real converts too. And the share of extra orders is not the same in every store. Big, save-up buys like furniture often test higher than cheap ones. The test shows the size of the leak. What you do next, limit it, talk the rate down, or keep it, is your call.
The sentence that changes how you think about BNPL
BNPL is not a payment method. It is a channel for winning new customers, and its fee is only worth paying on the customers it truly brings you.
Judge it the way you judge an ad channel. What does one truly new customer cost through BNPL? Is that cheaper than your other channels? The usage rate answers neither question.
2. How to run the BNPL incrementality test
This is a two-week numbers job. Use at least a full quarter of order history. You get two numbers out. The share of BNPL sales that are truly extra. And your cost per new customer.
- Split your BNPL orders into new and returning customers. These buyers already shop with you. BNPL is rarely why they said yes. Set them aside. Test the first-time buyers.
- Compare order values with and without BNPL. Look at new buyers of the same kind of product. Is the BNPL order value clearly higher? Maybe split payments let people buy the sofa, not the side table. That lift is real extra revenue.
- Test what happens to sales when BNPL is hidden. Run an A/B test that hides BNPL from a slice of traffic. Or compare the months before and after you added it. The gap in finished checkouts is your signal. Most BNPL companies publish their own lift data too. Use it as a cross-check only.
- Turn the signals into a working range. Combine the order-value lift and the checkout gap. Think ‘probably 25 to 40 percent extra.’ Not a fake-precise decimal. Small samples lie. That is why the test needs a full quarter of data.
- Price the new customers, then decide. Divide your yearly BNPL fees on extra orders by the count of extra buyers. Compare that cost to paid social and paid search. If BNPL wins, keep it or grow it. If it loses badly, limit it to big-ticket orders, where extra sales run highest. Or drop it. Re-test every year.
3. One warning before you act
Do not read this as ‘BNPL is bad.’ Think of slow, planned buys. Furniture. Mattresses. Jewelry. For buys like these, some stores find 60 percent or more of BNPL orders are truly extra. And $30 for a new customer beats almost any ad channel. The answer differs store by store. Run the test first.
And if the test says restrict or remove, move carefully. Customers who expect split payments notice when they vanish. A clumsy removal can cost more sales than the fees you save. Change one thing at a time. Watch your checkout numbers for a month. Be ready to bring BNPL back with limits. A minimum order size is the usual deal.
4. Frequently asked questions
A quarter of my orders use BNPL. Doesn’t that prove it works?
No. It proves shoppers like the button. Usage counts everyone. Even people who would have paid by card. Only one question matters for your money. How many of those sales would not exist without BNPL? That is what the test measures.
What share of extra sales is good enough to keep it?
There is no magic rate. Compare cost per new customer across your channels. Say paid social brings buyers at $80 each. Then $30 through BNPL is a happy deal. But say email and search bring them at $25. Then a $100 BNPL buyer looks bad.
What if BNPL customers come back and buy more over time?
Then the math improves. Maybe BNPL buyers show clearly higher lifetime value. Repeat orders. Bigger baskets. Then a higher first fee can be worth it. But track those buyers for a few months first. Assume nothing.
Should I offer several BNPL providers or just one?
Fewer is better. Pick one main provider. The one that sells best for you. That keeps the data clean. And it gives you volume to negotiate the rate. A wall of buttons splits your power without adding sales.
5. Quick reference: what to avoid and what to apply
What to avoid
- Judging BNPL by its usage rate, usage ignores whether the sale needed it.
- Paying a 6% fee on customers who would have happily paid by card.
- Drawing conclusions from 30 days of data or one product line, small samples lie.
- Ripping BNPL out overnight after one bad number, removal costs sales too.
- Comparing the BNPL fee to the card fee, the right compare is cost per new customer.
What to apply
- Split BNPL orders into new versus returning customers before anything else.
- Measure the order-value lift and the checkout lift BNPL really creates.
- Estimate a range of truly extra sales from at least a full quarter of data.
- Work out cost per new customer through BNPL and rank it against paid social and search.
- Restrict BNPL to high-value orders if the extra sales are thin, and re-test every year.
Want your six costs mapped for you?
Our free 31-Point Profit Diagnostic puts a dollar figure on where your profit is leaking. Or get a quote for done-for-you eCommerce accounting.
Definitions, modeling notes & rate-basis disclosures
Definitions
- The six profit levers
- In this series: (1) Discounts, (2) Refunds, (3) Product cost (landed COGS), (4) Sales channel and payment fees, (5) Shipping and 3PL fees, (6) Advertising spend.
- Buy Now Pay Later (BNPL)
- A checkout option that splits the customer’s payment into installments (usually four). Providers like Affirm, Klarna, and Afterpay charge the store 5 to 6 percent of the order for it.
- Incremental order
- A BNPL order that would not have happened, or would have been smaller, without the split payments. In plain words: an extra sale. A non-incremental order would have gone through on a card anyway.
- Incrementality rate
- The share of BNPL orders that are truly extra. Commonly 20 to 60 percent, depending on category and price.
- Customer Acquisition Cost (CAC)
- What a channel costs you per truly new customer. The honest yardstick for BNPL.
- Lifetime Value (LTV)
- The total revenue a customer brings over the whole relationship. High-LTV buyers can justify a higher cost to win them.
- Average Order Value (AOV)
- Total sales divided by number of orders. $500 in this article’s example.
Modeling notes
- The series’ standard teaching store sells at $100 with payment and channel fees near 3% = $3. This article differs on two counts, forced by the topic. The product is furniture at a $500 AOV (BNPL matters most on big-ticket orders). And the fee lever runs at the BNPL rate of 6% instead of 3%. On the $100 anchor that is $6 versus $3. On the $500 order, $30 versus about $15. The other levers hold at house proportions and are not shown.
- Per-order math: 6% of $500 = $30 BNPL fee.
- Yearly math: 5,000 BNPL orders × $30 = $150,000 total fees. At 30% incrementality: 1,500 extra orders × $30 = $45,000; 3,500 non-extra × $30 = $105,000. $45,000 + $105,000 = $150,000. Cost per new customer: $45,000 ÷ 1,500 = $30.
- The 30% incrementality figure is a teaching mid-range number. Real rates run roughly 15 to 70 percent by category. Measure your own before acting.
Rate-basis disclosures
- Baseline: furniture brand on Shopify, $500 AOV, 20,000 orders a year, of which 5,000 (25%) pay through BNPL.
- BNPL fee: 6% of order value, a working average across major providers (real rates run about 4 to 8 percent).
- Card benchmark: roughly 3%, about $15 on a $500 order versus $30 through BNPL.
- Incrementality: 30% assumed for the example; measure your own.
- All figures in whole dollars.