Payment & Processor Fees: The 2-3% That Breaks Break-Even
Most break-even models stop at gross margin. They subtract COGS, maybe shipping and fulfillment, and call the rest “contribution.” Then they divide to get a break-even ROAS and start judging campaigns against it. The problem: somewhere between the customer clicking “pay” and money landing in your account, the processor takes a cut you never modeled. It is small per order — in many cases 2-3% plus a fixed fee — but it comes off the top of every transaction, and it moves your real break-even point enough to flip “profitable” campaigns into quiet losers.
This is the most overlooked line in performance-marketing math: the relationship between payment processing fees and margin. Here is how the skim actually works, and how to fold it back into the numbers you optimize against.
For the neighboring economics, compare How a 20%-Off Code Silently Moves Your Break-Even ROAS and use The Hidden Fulfillment Costs That Reset Your Break-Even ROAS to validate the measurement decision.
The fee stack hiding under “contribution”
Every card transaction carries more than one charge. The headline number is the easy part. The full stack commonly includes:
- A percentage rate on transaction value (commonly published around 2.9% for blended flat-rate pricing).
- A fixed per-transaction fee (frequently a small flat amount per successful charge).
- Situational add-ons: premium-card surcharges, dispute fees, and sometimes payout or instant-payout fees.
Processors also price differently. Stripe and PayPal in many cases quote a blended flat rate: simple to forecast, slightly more expensive at scale. Adyen and Checkout.com more frequently run interchange-plus, which can be cheaper once you have volume but harder to predict. Same checkout, different effective take rate depending on which model you are on.
Treat these as illustrative published ranges, not fixed law — your real rate lives in your payout reports, not in a pricing page.
Why a “2.9%” rate is never actually 2.9%
The fixed per-transaction fee is what breaks the simple mental model, and it punishes low average order value (AOV) most difficult. A flat fee of roughly $0.30 is a rounding error on a large basket and a real tax on a small one.
| AOV | % rate | Fixed fee as % of order | Effective take rate |
|---|---|---|---|
| $120 | 2.9% | ~0.25% | ~3.15% |
| $60 | 2.9% | ~0.50% | ~3.40% |
| $25 | 2.9% | ~1.20% | ~4.10% |
A store with a $25 AOV is paying close to 4.1% in effective payment costs while believing it pays 2.9%. That 1.2-point gap is pure contribution margin, gone before any spend is recouped. The lower your AOV, the more the fixed fee dominates — which is exactly the segment where margins are already thin.
Refunds and chargebacks: the fees that compound
Returns and disputes are where the skim turns into a real leak, because the costs are asymmetric.
Refunds. When you refund an order, you do not always get the processing fee back. Depending on the processor and the era of your contract, you may lose the percentage fee, the fixed fee, or both. So a refunded order can cost you the fee twice — once when the sale processed, and again as an unrecovered charge — on top of the return shipping and restocking you already eat.
Chargebacks. A dispute is worse. You commonly pay a flat dispute fee per case (commonly cited in the mid-teens, varies by processor), and if you lose, you also forfeit the goods, the original shipping, and the revenue. Win the dispute and you may recover the amount but not always the fee. Card networks and processors also watch your dispute rate; cross a threshold — frequently framed around the 1% mark as an illustrative planning ceiling — and you risk monitoring programs, higher reserves, or punitive pricing. None of that shows up in a gross-margin model, but all of it lands on contribution.
The honest framing: a 1-2% refund-and-dispute drag on top of 3-4% base processing is entirely plausible for a typical store, and it stacks on the exact same revenue your ROAS target is built from.
How this resets your break-even ROAS
Here is the part that matters for media buying. Break-even ROAS is just the inverse of your contribution margin:
Break-even ROAS = 1 / contribution margin
Say your model shows 40% contribution before payment costs. Break-even ROAS = 1 / 0.40 = 2.50. Clean. Now fold in a realistic blended payment-and-dispute drag of 3% of revenue. Contribution drops from 40% to 37%. Break-even ROAS = 1 / 0.37 = 2.70.
That looks like a rounding error. It is not. Your true break-even just rose 8%. Every campaign sitting between 2.50 and 2.70 ROAS — the ones you have been scaling because they “clear break-even” — is actually underwater once the processor is paid. At low AOV, where the effective take rate runs 4%+, the gap is wider and the misjudgment is larger.
The same logic applies to MER (marketing efficiency ratio) at the account level. If you set a target MER off pre-fee contribution, you are systematically targeting a number that leaves no room for the cut every order already gave away. Payment processing fees compress margin quietly precisely because they never appear as a line in the ad platform — they live in a payout report you seldom open.
What to actually do about it
You do not fix this by switching processors on a hunch. You fix it by measuring the real number and putting it where decisions get made.
- Compute your true effective take rate. Pull a full month of payout/settlement reports. Divide total fees (processing + fixed + dispute + payout) by total processed revenue. That single percentage is your real payment cost — not the pricing-page rate.
- Fold it into contribution margin, then into break-even ROAS. Recompute break-even ROAS and target MER using the post-fee contribution. Publish the new break-even number to whoever buys media, because the old one is silently wrong.
- Segment by AOV. If you run a wide price ladder, a single blended break-even hides the truth. Low-AOV SKUs may need a meaningfully higher ROAS to clear; bundle or raise minimums to dilute the fixed fee.
- Attack the controllable leaks. Dispute fees and refund-fee loss are partly an operations problem: clearer billing descriptors, faster customer service, fraud screening, and prevention tooling cut the dispute rate before it touches pricing.
- Renegotiate or restructure once you have volume. At scale, interchange-plus can beat blended flat — but only model it against your actual card mix, not a quoted number.
This is exactly the kind of reconciliation that gets skipped because it lives between two systems no one owns. It is also where an operator-grade tool earns its place: Bach AI can read the real settlement-versus-revenue gap, recompute your true break-even ROAS, and flag the campaigns that only looked profitable — read-only, surfacing the leak for you to approve before anything changes.
The takeaway
Two to three percent sounds ignorable until you trace where it lands: skimmed before contribution is even counted, heaviest on low baskets, and compounded every time a refund or chargeback hits the same revenue your ROAS target is built from. Open the payout report, compute your true effective take rate, and rebuild break-even ROAS on post-fee contribution. The campaigns you scale should clear the real bar — not the one that pretends the processor works for free.