First-order profitability and CAC payback: how long until a new customer pays you back

The January customers looked like a win.

The ads worked. The orders landed. The team moved the budget up.

Months later somebody builds a duller table. It asks one question. Has that group of customers yet paid back what it cost to win them?

It has not.

The money went out in January, in one go. It comes back slowly, one reorder at a time, across the rest of the year. Nothing went wrong. That is simply how the arithmetic works when you pay for a customer up front.

But it is a fact you want before you raise the budget, not after.

First-order profit is what a new customer’s first order leaves after you subtract the cost of winning that customer. CAC payback period is how many days it takes a group of new customers to earn that cost back. Measure both on contribution margin, not revenue, and read payback on cohorts that are old enough to judge.

Payback is simplest on a single online store, where every order has a real customer email. If you also sell on Amazon, the first order has no customer ID you control. So the same question needs a different method there. Below: the four definitions, a worked example with illustrative numbers, the Amazon version, and the traps that make payback look faster than it is.

Four numbers, and one of them is allowed to be negative

CAC (customer acquisition cost) is what it costs in marketing to get one new customer. For a month, it is the marketing spend aimed at new customers divided by the number of new customers you won. A new customer is someone whose first order ever fell in that month.

First-order contribution margin is what the first order leaves after the costs that come with it. Start from what the customer paid, after discounts. Then subtract landed product cost (the product plus freight and duty to your warehouse), payment or marketplace fees, pick, pack and shipping, and expected refunds. Advertising stays out, because it sits on the CAC side. Our post on contribution margin for Amazon and Shopify builds this line by line.

First-order profit is first-order contribution minus CAC. It is often negative for products people buy again. That is not a failure. It is a bet that repeat orders will cover the gap.

CAC payback period is the number of days until a cohort’s cumulative contribution covers its CAC. A cohort is a group of customers whose first order fell in the same month. Cumulative contribution counts the first order plus every later order within the age you are measuring.

The formulas:

• First-order profit = first-order contribution margin − CAC
• Payback reached at age N when cumulative contribution per customer at N days ≥ CAC

A worked example: a cohort of 1,000 new customers

Illustrative numbers only. A brand sells hand soap on its own online store. In March it wins 1,000 new customers and spends $38,000 on marketing to do it. CAC is $38.

The average first order is $60 after discounts. After product cost, payment fees, shipping and expected refunds, it leaves $21 of contribution margin. So first-order profit is $21 − $38 = −$17 per customer. The brand is $17,000 in the hole on day one.

Then some customers come back. Each cell below is cumulative contribution per customer, first order included.

At 90 days the cohort has earned back 87% of what it cost. By 180 days it has earned back all of it, plus $5 per customer. Payback lands between 90 and 180 days. A straight line between those two readings puts it near day 135. That is an estimate between two readings, not a measured day. Adding readings at 120 and 150 days would pin it down.

Two things the table does not tell you. It does not say whether a payback near 135 days is good. That depends on how long your cash can wait, which is a choice only you can make. And it says nothing about the April cohort, which may behave differently. Build the same table for every month.

The Amazon version: estimate, and say what is estimated

On your own store, a customer has a real email from the first order. You know who is new. On Amazon you do not.

Amazon’s order data gives you an anonymized buyer email, an alias Amazon creates, rather than the shopper’s real address (Orders API model). You can use it to link one buyer’s Amazon orders with you over time. You cannot match it to your store, your email list or an ad platform. In the data we see, it also fills in some days after the order, and on some brands it is missing for whole months. Our post on Amazon customer lifetime value covers how to check for that.

So on Amazon, both sides of payback are built with care.

Two ways to count a new customer, and they disagree

There are two ways, and they measure different things.

• From Amazon Ads. Amazon Ads counts an order as new-to-brand when the shopper has not bought from your brand on Amazon in the past 12 months (Amazon Ads new-to-brand metrics). These are ad-attributed orders only, not every new customer. And Amazon says new-to-brand data covers Sponsored Brands and Sponsored Display, not Sponsored Products (the ads that promote one product in search results). In the ad accounts we see, Sponsored Products is often most of the spend.
• From your order data. Count buyers whose first order ever, across all the history you hold, fell in the month. This covers every new buyer, from ads or not. It is only as complete as the buyer alias is for those months.

The two definitions do not match. Twelve months without a purchase is not the same as never having bought. Never divide spend from one by customers from the other and call it one number.

The Amazon CAC estimate, and what it gets wrong

The usual Amazon estimate is total Amazon ad spend divided by first-time buyers from your order data. Say plainly what that estimate does:

• It counts ad spend that also reached repeat buyers, which pushes CAC up.
• It credits ads with new buyers who found you through search on their own, which pushes CAC down.
• It undercounts customers in any month where buyer aliases are missing, which pushes CAC up.

You cannot remove these from Amazon data. You can keep the method fixed, write it down, and label the result “estimated CAC.” A consistent estimate shows a trend. A method that changes every month shows nothing.

What is measured and what is not

Contribution per order on Amazon is the most solid piece. Referral fees (Amazon’s commission on each sale), FBA fees (Amazon’s charge to pick, pack and ship) and refunds come from the settlement report. Amazon does not know what your product cost you. That part comes from the costs you enter yourself, so it is only as good as those.

Where it goes wrong

These are the traps we see most often in real multi-channel data.

Returning customers counted as new

A “new customer” who last bought two years ago is not new. But a tool that calls anyone new who had no order inside the window you picked will count them. So will a report that starts its history a year ago.

The effect is large. Say 200 of the 1,000 “new” customers in the example had bought before. The real new count is 800, and real CAC is $38,000 ÷ 800 = $47.50. Even the $43 reading at 180 days falls $4.50 short, so payback moves past 180 days. Worse, returning customers reorder more. Taking them out lowers the contribution line too, which pushes payback later still.

Blended CAC hiding a channel that never pays back

Blended CAC puts every channel’s spend over every channel’s customers. It can hide a channel that loses money.

Illustrative numbers. The blend pays back inside 180 days. Channel B does not. Its customers had earned back only $37 of a $50 cost by day 180. Work out CAC and payback per channel, then look at the blend.

Payback on revenue happens on day one and means nothing

In the example, the first order is $60 and CAC is $38. On revenue, payback happens on day one. On contribution, it takes an estimated 135 days. The difference is product cost, fees and shipping, money the brand never keeps. Payback built on revenue makes almost every channel look like it pays for itself. The same logic sits behind break-even ROAS and MER. ROAS is sales divided by ad spend, and MER is total sales divided by total marketing spend. Both also need margin, not revenue.

A cohort that is not slow, only unfinished

On September 15, a June cohort has not fully reached 90 days. A customer who first bought on June 30 reaches 90 days on September 28. If you fill in the June 90-day cell today, some customers have had far less time to reorder. That cohort looks like it will never pay back. It is not slow. It is unfinished. Leave a cell blank until every customer in the cohort has reached that age.

A second version of this trap: CAC from this month set next to contribution from a cohort won a year ago. Match the months. The CAC and the cohort must come from the same period.

A welcome discount digs the hole deeper

A 30% welcome discount on a $60 first order takes $18 off the price. The first-order contribution falls from $21 to about $3. First-order profit falls from −$17 to about −$35.

That is the visible cost. The cohort starts $18 deeper in the hole. A discount may also draw shoppers who came for the price and come back less at full price. Check that in the cohort’s later cells rather than assume it either way. Read a discount month’s cohort on its own, and do not let a strong full-price cohort average it away.

How to set it up

1. Write down what “new customer” means, once. First order ever, across all the history you hold, per channel. On Amazon, say whether you use the buyer alias or Amazon Ads new-to-brand. Every report uses the same definition.
2. Work out CAC per channel, per month. Put spend and new customers from the same month side by side. On Amazon, label it estimated and write down the method.
3. Attach contribution margin to every order. Subtract discounts, landed product cost, fees, shipping and refunds. Leave advertising out.
4. Build the cohort table at fixed ages. Cumulative contribution per customer at the first order, 30, 60, 90 and 180 days. Add 365 days once you have the history. Leave a cell blank until the whole cohort has reached that age.
5. Mark the payback age per cohort and per channel. The first age where cumulative contribution reaches CAC. If no filled cell reaches it yet, write “not yet paid back at N days.”
6. Set a payback target yourself. Pick it from your cash position and how long you can wait, not from an industry number. Write down the age it applies to, for example “paid back within 120 days.”
7. Watch the trend, not one cohort. Are new cohorts paying back sooner or later than older ones at the same age? That is the question that changes a budget.

Shopify’s customer cohort analysis report groups customers by the date of their first order (Shopify customers reports). It shows sales and retention, not margin. That makes it a good place to check your customer counts. The contribution side still has to come from your own costs.

So before the next budget meeting, build one cohort table and write your own payback target above it. Then the question stops being whether the ads worked, and becomes whether the money comes back in time.

Synthesis keeps a brand’s definitions and its own targets written down once, and every answer shares them. Customer numbers carry their caveats, such as “this cohort is too young to judge repeat rate” and “Amazon buyer emails arrive late so recent customer counts run low.”