Target CPA: how to work out the cost per acquisition you can afford, and which one to bid against
Your target CPA is the cost per acquisition you aim for, set below the most you can afford to pay. This post works out that ceiling three ways (first order, 60 days and lifetime), shows which one to bid against based on your cash, and explains how it fits with Meta's and Google's cost goals.
On this page
- How to calculate your allowable CPA from the first order
- Allowable CPA over 60 days and over a customer's lifetime
- Which CPA should you bid against? Follow the cash
- How to set your target CPA: a worked example
- Target CPA in Meta and Google vs your allowable CPA
- Target CPA or target ROAS: which should you use?
- When your CPA looks fine but you're losing money
- What to do this week
- Frequently asked questions
A target CPA is the cost per acquisition you aim to pay for a new customer, set below your allowable CPA, the most you can pay without losing money. On the first order, allowable CPA equals that order's contribution: average order value × contribution margin. Over 60 days or a lifetime, multiply by the orders a customer places in that time. Then bid against the horizon your cash can carry.
How to calculate your allowable CPA from the first order
Allowable CPA, also called allowable acquisition cost or maximum CPA, is the ceiling: what one new customer can cost before the order that brought them loses money. On the first order it's simply the contribution that order leaves, everything you keep after the costs that come with it.
Allowable CPA and target CPA
Allowable CPA, first orderaverage order value × contribution margin
Allowable CPA, any horizoncontribution per order × orders per customer in that time
Target CPAallowable CPA − the profit you want per customer
Illustrative: $80 × 55% = $44.00. Over 60 days, $44 × 1.3 orders = $57.20. With $8 of profit per customer, target CPA = $49.20.
Take an illustrative supplement brand whose average order, bundles and checkout upsells included, is $80. Each order costs $17 in product and $9 to ship and fulfill. Payment fees are 4%, refunds run at 8% of revenue, and chargebacks plus apps billed per order add $0.40 when spread across every order. That leaves $44.00, a 55% contribution margin:
From an $80 order to the first-order allowable CPA
Illustrative example, per order
So the first-order allowable CPA is $44. Pay $44 for a new customer and the first order breaks even. Pay $50 and you're $6 down on that customer until they order again. If your contribution margin is new to you, the break-even ROAS post builds it step by step with a calculator; the break-even CPA there is the same number as the first-order allowable CPA here.
Count upsells at their take rate
The $80 average order already includes the upsell. Front-end orders average $68, and 30% of buyers add a $40 upsell right after checkout, so each customer brings 0.30 × $40 = $12 of upsell revenue on average: $68 + $12 = $80. Count an upsell at its take rate, never at its full price, and use its own costs: an upsell that ships in the same box adds product cost but no second postage.
Allowable CPA over 60 days and over a customer's lifetime
Supplements run out, and some buyers come back. Every repeat order brings another $44 of contribution with no ad cost attached, so a customer is worth more the longer you wait to count. The catch: you can only count what you've seen.
Take a cohort, every customer whose first order fell in one past month, and count their orders. Say that by day 60, 1.3 orders per customer have come in (30 of every 100 buyers reordered once), and by day 365, 2.2. The same customer now supports three different ceilings:
Allowable CPA by horizon
Illustrative example: $44 of contribution per order
Those numbers only hold for cohorts old enough to have reached that day. A month-old cohort tells you nothing about day 365, and a launch with no repeat buyers yet has only one honest number: the first order. If your repeat orders are smaller than first orders, a single bottle instead of a bundle, use their own contribution rather than $44. Upsells sold after checkout, or by email in the following weeks, count in the horizon they arrive in.
Which CPA should you bid against? Follow the cash
The three ceilings answer different questions, and the difference between them is cash. Ad spend leaves your account within days. Repeat orders arrive over months.
Say the brand bids up to its 60-day ceiling, about $57. Each new customer costs $57 and returns $44 on the first order, so $13 per customer sits in the gap until the reorders come in. At 300 new customers a week, that's $3,900 a week. With reorders spread over 60 days, up to 60 days of that gap, about 8.6 weeks or roughly $33,400, can be out of your account at once. Fine if you have it; a problem if you don't.
Choose the horizon with three rules:
- Bid against the first order when cash is tight, the offer is new, or you have no repeat data. Nothing waits on future orders.
- Bid against 30 to 60 days when you have at least one complete cohort that old and the cash to carry the gap.
- Use lifetime value only with a payback rule. The 12-month ceiling says $96.80, but if your rule is that every customer must pay back within 60 days, your ceiling is $57.20. Take the lower of the two.
Payback is the same idea counted in orders: CPA ÷ contribution per order. A $52 CPA ÷ $44 = 1.18 orders, so that customer pays back partway into their second order. The LTV to CAC post shows how to read the payback month off a cohort and how much cash it ties up:
How to set your target CPA: a worked example
Allowable CPA is where profit hits zero. Target CPA leaves room for the profit you want: target CPA = allowable CPA at your chosen horizon − profit per customer.
The brand bids against 60 days and wants $8 of profit per customer by day 60, so its target CPA is $57.20 − $8 = $49.20. Its campaign is running at a $52 CPA: above target, below the ceiling. The calculator below starts with that campaign and the brand's 12-month cohort (2.2 orders per customer). Its 55% margin already includes refunds and chargebacks.
Calculator
LTV to CAC calculator
What a customer is worth over all their orders, against what it costs to win them.
- Lifetime profit per customer (before ads)
- First order after paying for the customer
- Orders it takes to earn the CAC back
LTV here is lifetime contribution — revenue minus product, shipping and fees — not revenue. Use a lifetime you have actually observed, not one you hope for.
Read the results against the three horizons. The first order loses $8. By day 60, the customer has brought $57.20 and the brand is $5.20 ahead, short of its $8 target. By month 12, the observed 2.2 orders bring $96.80, which is $44.80 more than the CPA.
To close the $2.80 gap, the brand can bring its CPA down to $49.20, or raise what each order leaves. If a better bundle mix lifts the average order from $80 to $85 at the same 55% margin, each order leaves $46.75 and the 60-day value becomes $60.78. The target rises to $52.78, and the current $52 CPA meets it. The average order value post lists the levers. The price itself sets what each order leaves; our post on how to price supplements builds a price floor from an order's ad cost.
Target CPA in Meta and Google vs your allowable CPA
The ad platforms ask for a cost target too, but theirs is a bidding instruction, not a promise.
- Meta. In Ads Manager the number goes in the Cost per result goal field, which Meta's own guide describes as the bid you're willing to pay for your performance goal (Meta). Meta's API describes a cost cap as getting as many results as possible while it tries to meet the cost per action you set, and says adherence isn't guaranteed. A bid cap, by contrast, raises your bid as needed but never above your limit (Meta bid strategies).
- Google. Target CPA bidding sets bids to get as many conversions as possible around the target you set, and Google's help page says plainly: "Some conversions may cost more than your target and some may cost less" (Google Ads Help).
- Delivery. The same Meta guide warns that too low a bid may keep an ad set from collecting the minimum 50 events over 7 days it needs to exit the learning phase. The learning phase post covers what that means for budgets.
Three rules follow:
- Give the platform your target, not your ceiling. Averages drift, so a goal at $49.20 leaves room for days that run high before you reach the $57.20 break-even.
- Watch whether it spends. If an ad set with a cost goal can't spend its budget, the goal is probably below what the auction clears. Raise it toward your target in steps, never past it.
- Check the platform's CPA against your store. The platform divides spend by the purchases it attributes to its own ads, which can include returning customers. Your allowable CPA is about new customers and real, net orders.
How you split campaigns and budgets around these goals is its own decision; the CBO vs ABO guide covers it.
Target CPA or target ROAS: which should you use?
The two goals are the same economics in different units. Target ROAS = average order value ÷ target CPA, so the brand's $49.20 target CPA on an $80 average order is a target ROAS of $80 ÷ $49.20 = 1.63x.
The difference is what the platform optimizes. A CPA goal treats every purchase as equal. A value goal weighs each purchase by its size: Google's Target ROAS, for example, predicts the value of a potential conversion when it sets a bid (Google Ads Help). A rule of thumb follows:
- Most orders close to one price (a single offer, little bundle spread): a CPA goal is simpler and says exactly what you can pay.
- Order values spread widely (one, three and six bottles, from $69 to $294): a value goal lets the platform chase the bigger orders, but only if your purchase events pass the real order value.
Whichever you pick, derive it from the same allowable CPA, so the two can never disagree about what a customer is worth.
When your CPA looks fine but you're losing money
A campaign can hit its CPA every day and still lose money, because the costs that decide profit arrive after the ads are paid for. Here is the same brand with a plan built on a 3% refund rate and no allowance for chargebacks. On paper each order left $48.40, so a $46 CPA looked like $2.40 of profit per order. Then the month closed:
The plan vs the month's real numbers
Illustrative example, per order
On paperThe plan
- Refunds: $2.40 (3%)
- Chargebacks and per-order apps: $0.00
- Contribution: $48.40
- Ad cost per order (CPA): $46.00
- Result per order: +$2.40
After the month closedThe month's real numbers
- Refunds: $6.40 (8%)
- Chargebacks and per-order apps: $0.40
- Contribution: $44.00
- Ad cost per order (CPA): $46.00
- Result per order: −$2.00
Across 1,000 orders, that's a $2,000 loss where the plan promised $2,400 of profit: a $4,400 swing, while the ads did exactly what they were told. The usual causes:
- Refunds arrive late. A new campaign's refund rate looks low because the refunds haven't come in yet. Measure it on orders older than your guarantee period.
- Chargebacks cost more than the order. You lose the disputed amount plus a fee. Stripe charges US businesses $15 per dispute, and for disputes opened after June 17, 2025, another $15 if you counter, returned only if you win (Stripe pricing). Cardholders can typically dispute within 120 days of the payment, sometimes longer (how disputes work). The chargeback reduction post covers prevention.
- Fulfillment errors. A lost parcel, a damaged bottle or a wrong item means paying for the product and the shipping a second time.
- The wrong denominator. If the platform's CPA counts returning customers, your true cost per new customer is higher than it shows.
The fix is a monthly reconciliation: real costs per order, real refund and chargeback rates, real new customers. Profit tracking in RevvyScales (Growth plan and up) does the subtraction for you: ad spend from a connected Meta account, product costs and gateway fees, with ROAS and margin by campaign (profit tracking). Whatever tool you use, check its refund and chargeback numbers against your processor's reports each month.
What to do this week
- Work out first-order contribution from last month's real costs, refunds and chargebacks included.
- Pull a cohort at least 60 days old and count orders per customer by day 60, and by day 365 if you have a cohort that old.
- Write down your three allowable CPAs and choose the horizon your cash can carry.
- Subtract the profit you want per customer. That's your target CPA; put it in the platform's cost goal instead of the allowable.
- Compare last week's CPA, counted from your store's new customers, with both numbers.
- Recheck refunds and chargebacks every month and update the margin when they move.
Frequently asked questions
What is the difference between target CPA and allowable CPA?
Allowable CPA is the most you can pay for a customer without losing money over a chosen horizon: contribution per order times the orders a customer places in that time. Target CPA is the number you aim for, set below the allowable by the profit you want per customer. Give the platform the target, so ordinary swings above it still land below break-even.
How do you calculate CPA?
CPA, cost per acquisition, is ad spend divided by the conversions it produced, usually purchases or new customers. $5,000 of spend that brings 100 orders is a $50 CPA. For profit decisions, count conversions from your own store data rather than only the ad platform report, and decide whether you mean every purchase or new customers only, since returning buyers make CPA look cheaper.
Should I use lifetime value to set my target CPA?
Only with real data and a payback rule. Lifetime value says what a customer can be worth, but ad spend leaves your account within days while repeat orders arrive over months. Use the orders per customer you have actually observed in an old enough cohort, and cap your CPA at what comes back within the time your cash can carry, for example 60 days.
Why is my CPA in Ads Manager lower than in my store?
The platform counts the purchases it attributes to its own ads under its own rules. That can include sales another channel also claims, purchases by returning customers and orders that are later refunded. Your store knows which orders were real and which were refunded. Compare both for the same days, and set your target CPA against the store figure.
What happens if I set my cost per result goal too low?
Delivery can slow or stall, because the platform cannot win enough auctions at that price. Meta warns in its own guide that too low a bid may keep an ad set from collecting the minimum 50 events over 7 days it needs to exit the learning phase. Start the goal at your target CPA, watch whether the ad set spends, and lower it only if results hold.
Is a lower CPA always better?
No. A cheaper CPA can come from smaller orders, buyers who refund more, or returning customers counted as new. Judge CPA next to the contribution each customer brings: a $60 CPA on customers worth $80 in 60 days beats a $40 CPA on customers worth $45. Profit per customer, not CPA alone, decides what to scale.