Most teams treat the checkout as a design surface. They argue about button color and field order, ship a redesign, and call it a win when bounce rate dips. That framing is the reason so much revenue leaks out of the bottom of the funnel without anyone owning the loss.
A checkout is an economic system. Money enters at the top as intent and exits at the bottom as settled revenue, and at every stage between, some fraction is lost to friction, declines, fees, and failed retries. If you cannot read your checkout as a profit and loss statement, you cannot manage it. You are just decorating the place where the money escapes.
The four levers
Every dollar that reaches the buy button passes through four gates. Each one is a percentage, and they multiply.
1. Conversion
This is the share of shoppers who reach the checkout and actually submit a payment. It captures form friction, trust signals, payment method availability, and load time. It is the lever product and design teams already obsess over, and it is real, but it is only the first multiplier.
2. Authorization
Once a shopper submits, the transaction has to be approved by the issuer. Approval is not guaranteed. Global card authorization rates typically sit between 85 and 95 percent, and card-not-present ecommerce runs roughly 10 percentage points below in-person rates. A large share of those declines are false positives on legitimate buyers, not fraud. We cover the mechanics in Module 2; here, the point is that authorization is a conversion event you do not control from the front end.
3. Cost
Every approved transaction carries a cost of acceptance. Interchange goes to the issuer and is the largest line, often 75 to 85 percent of total processing cost, and averaging around 2 percent of transaction value on US credit. Network assessments add roughly 0.11 to 0.13 percent, and the acquirer or processor takes a margin on top. Cost does not change whether the sale happens, but it changes how much of the sale you keep.
4. Recovery
Some failures are recoverable. A soft decline can be retried, an expired card can be updated, a failed subscription renewal can be dunned. Recovery is the lever most teams never instrument at all, which means they write off revenue that was one retry away from settling. Module 8 goes deep on it.
Why you have to model them together
The trap is optimizing one lever in isolation. A team adds an aggressive fraud rule and watches chargebacks fall, then never notices that authorization quietly dropped two points because legitimate buyers got blocked. A team chases least-cost routing and shaves fees, then loses more in declined volume than it saved in basis points.
Because the levers multiply, a small move in one can be erased by a smaller move in another. The only honest way to evaluate a checkout change is against settled revenue net of cost, not against any single rate.
A worked example
Take a merchant processing 100,000 checkout attempts a month at a $100 average order value.
- Conversion to submitted payment: 70 percent, so 70,000 attempts.
- Authorization rate: 87 percent, so 60,900 approved transactions.
- Gross approved revenue: $6,090,000.
- Cost of acceptance at 2.3 percent all-in: about $140,000.
- Net revenue: roughly $5,950,000.
Now suppose we lift authorization from 87 to 90 percent through better retry logic and routing, with no change to traffic or conversion. Approved transactions rise to 63,000, gross approved revenue to $6,300,000. That is $210,000 in additional monthly revenue from three points of authorization, against a marginal increase in processing cost of a few thousand dollars.
Compare that to a conversion redesign that lifts submission from 70 to 71 percent. That is worth about $87,000 a month at the same authorization and cost. Both are real wins. But a team that only watches the front end would have spent its quarter on the smaller of the two, because authorization never showed up on its dashboard.
The one number that matters
The metric that ties the four levers together is revenue per checkout attempt, net of acceptance cost. In the example above that is roughly $59.50. It is the only figure that moves correctly no matter which lever you pull, because it already accounts for the multiplication. Conversion, authorization, cost, and recovery are the inputs; net revenue per attempt is the output you actually manage.
When you frame it this way, the org chart usually has a problem. Conversion belongs to product, authorization to payments, cost to finance, recovery to nobody. The economic system spans four teams, and the gaps between them are exactly where revenue leaks.
Takeaway
Stop asking whether the checkout looks good and start asking what it earns per attempt. Build a single model with conversion, authorization, cost, and recovery as multiplying inputs, hold every change accountable to net revenue per checkout attempt, and give one owner the whole equation. The rest of this course is about working each lever, but they only pay off if you read them on one statement.