Does Your Ad Math Add Up? Break-Even ROAS, MER, and How Much a Test Really Costs

Ads Manager says your ROAS is 2.6. That sounds like you turn every dollar into $2.60, so you’ve been thinking about doubling the budget. But when you look at your bank balance at the end of the month, it hasn’t moved much, and you can’t tell whether the ads are making money or just costing it.

That gap is almost never a Meta problem. It’s an arithmetic problem: nobody worked out what a “good” ROAS is for this product, and nobody checked the number Meta reports against the money that actually came in. This page walks one made-up store through the whole calculation, from the first cost line to the decision of whether to scale.

Everything below is a hypothetical example. The store, the prices and the results are invented to show the arithmetic, not taken from a real customer. Replace every number with your own.

The example store: a Shopify store selling a dog joint supplement at $40 a tub, free shipping. It runs Meta ads, has a Klaviyo flow, and some customers already reorder.

Number 1: what one order actually earns you

Start with the price and take away everything that only happens because that order happened. (Fixed costs like your Shopify plan, apps and your own salary come later, from the weekly profit, not from each order.)

A few things people leave out:

• International cards cost more. Shopify’s pricing page lists an extra 1% for international cards on every plan. If a quarter of your orders come from Canada or the UK, use a blended rate.
• If you don’t use Shopify Payments, Shopify adds a 2% transaction fee on Basic on top of whatever your payment provider charges (same pricing page).
• Returns are not a rounding error in every category. The National Retail Federation and Happy Returns estimated that retailers expected 19.3% of online purchases to be returned in 2025 (NRF, 2025). That’s an average across large U.S. retailers; a supplement will run far lower, shoes higher. Use your own Shopify refund report.

The $19.54 is your contribution per order. Divide it by the price and you get your contribution margin: 19.54 ÷ 40 = 48.9%.

Number 2: break-even ROAS

Break-even ROAS is the point where ads pay for themselves and nothing more:

Break-even ROAS = 1 ÷ contribution margin

For the example store: 1 ÷ 0.489 = 2.05. Every dollar of ads has to bring back $2.05 in sales just to get back to zero.

This turns “ROAS 2.6” into a verdict. At 2.6 this store makes money on each attributed order. The same $40 item at a 30% margin needs 3.3 to break even, and there the same 2.6 loses money on every sale Meta reports.

Number 3: the most you can pay for a customer

Most owners think more clearly in dollars, so flip it:

Break-even cost per purchase = contribution per order = $19.54

If Meta’s cost per purchase in Ads Manager is above $19.54, the first order loses money. Then decide how much profit you actually want from that first order. Say you want $6 left over: your target cost per purchase is $13.54, which is a target ROAS of 40 ÷ 13.54 = 2.95.

Write those three numbers on a sticky note: $19.54 profit per order, 2.05 break-even ROAS, $13.54 target cost per purchase. Everything else on this page is about checking them against reality.

Why the ROAS in Ads Manager isn’t the money in your account

Meta only reports the purchases it can link to an ad, inside a time window. Its standard attribution setting counts a purchase if it happens within 7 days of someone clicking an ad or 1 day after they viewed it (see Meta’s About attribution settings and Jon Loomer’s guide to the attribution setting). Two things follow:

1. Some of those purchases would have happened anyway. Someone who already gets your emails scrolls past an ad and buys the next day from a Klaviyo email. Meta counts it under the 1-day view window. Klaviyo also counts it. So does Google, if they searched your brand name first.
2. Add up every platform’s “conversions” and you’ll often get more orders than Shopify actually has. Each tool is claiming the same order.

None of this means Meta is lying. It’s answering a narrower question (“which orders touched an ad?”) than the one you care about (“how much extra money did the ads bring in?”).

The number that doesn’t double count: MER

MER (marketing efficiency ratio) = total store revenue ÷ total ad spend, across every channel, from Shopify’s sales report and your ad invoices. It can’t double count because there’s one revenue number and one spend number.

Here’s one week in the example store:

ROAS 2.6, MER 3.0. Looks great on both counts. Now the question nobody asks: what did the store sell in weeks when the ads were off? Say the answer is about $1,500 a week, from repeat customers, email and search. Then:

• Extra revenue the ads plausibly drove: 4,200 − 1,500 = $2,700
• Real return on the ads: 2,700 ÷ 1,400 = 1.93, below the 2.05 break-even

In cash: the week earned 48.9% × $4,200 = $2,052 in contribution, minus $1,400 of ads = $652. A week with no ads would have earned 48.9% × $1,500 = $733. The ads lost about $80 that week while Ads Manager showed 2.6.

Your baseline won’t be that clean, but a rough “what happens without ads” number from a quiet week or two is enough.

Watch weekly cash, not daily ROAS

Once a week, write down five numbers: total revenue, total ad spend, MER, new customers (Shopify’s reports can split first-time and returning customers), and cash in the bank. Daily ROAS swings with which day a big order happened to land. Weekly totals, compared to your baseline, tell you whether the account is growing.

When the first order loses money: repeat purchases

Supplements, coffee, skincare, pet food: in these categories, the first order losing money can be a deliberate choice, if customers come back. Here’s the rough version.

Assume (and this has to be an assumption until you have data) that in the first 12 months the average new customer places 1.6 orders, and repeat orders cost almost nothing to win because they come from email.

• 12-month contribution per customer: 1.6 × $19.54 = $31.26
• So over a year, you could pay up to about $31 for a customer and still break even.

Three cautions before you raise your target cost per purchase to $31:

1. Use your real repeat rate, not a hopeful one. Shopify Analytics has a customer cohort report. Wait until you have at least 90 days of cohorts before you trust it, and use the worst recent month.
2. You pay for the customer today and earn the money back over months. If you pay $28 per customer and add 300 customers a month, you’re $8.46 short on each first order, about $2,500 a month that you have to cover until the reorders come in. Check that against your bank balance, not just the spreadsheet.
3. Use a shorter window when cash is tight. Use 90-day value, not 12-month value, to set your bid until the business can survive the wait.

When to add money, when to cut

Pick the rules before you look at the results, so you can’t talk yourself into them afterward. A reasonable starting set:

• Scale when the last 7 days’ cost per new customer is under your target and the weekly cash number is rising. Raise budgets in steps (a common habit is 20–30% at a time) rather than doubling; big jumps change who Meta shows your ads to, and the next week’s numbers won’t compare cleanly.
• Hold when you’re between target and break-even. You’re not losing money; work on creative instead of budget.
• Cut when cost per new customer has been above break-even (or above your 90-day value, if you sell on repeat) for two weeks in a row.

How much does a test have to cost before it tells you anything?

Here’s what quietly ruins most small-budget testing: you spend $30 on a new creative, get one sale, and call it a winner, or get zero and call it dead. Neither conclusion is supported. The budget has to be big enough to produce the number of events you want to read.

Test budget = number of events you need × what each event costs

Assume for the example store that a link click costs $0.70 and 3.5% of visitors buy. That makes cost per purchase about $20, right at break-even.

Stage 1: is the creative stopping people? Read click-level metrics like cost per link click and click-through rate. They come in fast. Aim for roughly 150 link clicks per creative: 150 × $0.70 ≈ $105. Use this to drop the obvious losers, not to declare a winner.

Stage 2: does it sell at a price you can live with? For that you need purchases, and purchases are rare, so the noise is big. Here’s how wide the honest range is, using standard Poisson confidence intervals:

Three sales tell you almost nothing. Ten tells you whether you’re in the right range. Thirty is when you can tell a $16 creative from a $24 one. So a real purchase test for one creative in one country costs roughly 10 purchases × your expected cost per purchase to start, and about three times that to decide between close options.

When to stop early: if a creative has spent 3× your break-even cost per purchase (here about $60) with zero sales, stop it. If its true cost were at break-even, you’d expect about 3 sales by then, and the chance of seeing none is only about 5%.

Two more rules that save money:

• Test one thing at a time. One country, one offer, several creatives, or one creative across two countries. If you change the creative and the country together and it wins, you won’t know why.
• Count the cost of the creative itself. If each new video costs you money or a day of work, that’s part of the test budget. Five creatives at $105 each in Stage 1 is $525 in ad spend plus whatever the five creatives cost to make.

The whole thing on one page

1. Profit per order = price − product − shipping − payment fees − refunds → $19.54. Break-even ROAS = 1 ÷ margin → 2.05. Target cost per purchase → $13.54.
2. Every week: MER, revenue against your no-ads baseline, new customers, cash.
3. Repeat business justifies a higher cost per customer only at your measured repeat rate, and only if you can carry the gap.
4. Tests: ~150 clicks to judge the hook, ~10 purchases to judge the price, stop at 3× break-even with zero sales.

What to do next

Open a spreadsheet tonight and fill in the five cost lines for your best-selling product. If your break-even ROAS comes out higher than the ROAS you’re getting, fix that before you touch the budget.

Then plan your first properly sized test: How to Test Ad Creative covers what to vary, and Why Your Ads Get Junk Leads covers what to do once one creative wins.

If you want a fixed number for the creative-cost line: in AutoWhisper one full video is 52 credits (what credits cost depends on your plan; see the pricing page). The ads it builds in Meta start paused. You decide when to turn them on and how much to spend.