A 2.0 ROAS at a 40% margin
The ad dashboard calls it a return. Your bank account calls it a loss. The arithmetic follows, with how agents judge spend by what each sale is worth to you.
Heraldic · Written
Every ad platform reports a return on ad spend, or ROAS: the revenue it credits to your ads, divided by what you paid for them. A ROAS of 2.0 sounds like doubling your money. For a business that sells physical products, it usually isn’t.
Take a store that spends $1,000 on ads in a week and has a 40% gross margin. Of every dollar of revenue, 40 cents is left after paying for the goods.
The dashboard shows $2,000 back on $1,000. The business is $200 poorer.
Break-even is one divided by your margin
Ads pay for themselves when the gross profit they bring in equals what they cost. Gross profit is revenue times margin, and revenue is spend times ROAS, so the break-even ROAS is one divided by the margin.
| ROAS | Revenue | Gross profit | After ad spend |
|---|---|---|---|
| 1.5 | $1,500 | $600 | −$400 |
| 2.0 | $2,000 | $800 | −$200 |
| 2.5 | $2,500 | $1,000 | $0 |
| 3.0 | $3,000 | $1,200 | +$200 |
| 4.0 | $4,000 | $1,600 | +$600 |
Your real margin is usually thinner
Gross margin leaves out costs that grow with every order: card fees, shipping you absorb, packaging, returns and discounts. Say card fees take 3% of revenue and free shipping takes another 5%. That leaves a 32% contribution margin to pay for the ads, and the break-even point rises.
Why the dashboard still looks fine
Three things make a losing week look healthy. First, ROAS counts revenue, not profit, and the platform doesn’t know your margin unless you tell it. Second, each platform counts its own conversions in its own attribution window. A customer who saw an ad on one platform and clicked an ad on another can be claimed by both, so together the platforms can report more orders than your store actually took. Third, reconciling the two takes time that a busy team rarely has.
The platforms deserve credit. The big ones let you send a value with each conversion and bid toward it, and if you send margin instead of revenue, their bidding gets closer to profit. That is a real improvement. But each platform still grades itself on its own count, sees only its own campaigns, and can’t see your margin by product line, your email or what you spend elsewhere.
Judging each sale by what it’s worth
Agents don’t judge themselves on a platform’s ROAS. Each agent has a funnel: the steps people take toward a result, from an ad click to an order, and what each step is worth to you. A step’s worth is either a fixed amount, like $8 for a started checkout, or a value read from the event itself, such as the order value or the margin on that order. Agents compare what they spend with what the results are worth.
Each brief can also set a target on one step of the funnel, such as first orders under a set price. And agents read your business signals, such as stock levels and margin tables from your warehouse, as context when they plan.
The results come from your own conversion events, sent from your store, your CRM or the events API. Each platform’s own numbers still steer that platform’s bidding. Decisions across channels, such as moving money from one platform to another, are judged on your events.
Agents judge each sale by what it’s worth to you: order value, margin, or any value you send with the event.
A coffee roaster’s numbers, at an example margin
Wrenfield Coffee, the roaster in the live demo (opens in a new tab), has a New Customers agent with a funnel in which a first order is worth its order value, about $47 on average, and a new subscription is worth its first-year value, about $312. Its brief sets a target of first orders under $26. Its latest report: 71 first orders at $20 each, and 16 of those 71 new customers started a subscription.
The roaster’s funnel values orders at revenue. Apply this essay’s 40% margin as an example, and the target looks like a loss on the first order alone:
A target above the profit on a first order is a bet on what comes after it. This week, about one new customer in four and a half started a subscription. Counting the subscription on top of the first order, at the same example margin:
- First-year value isn’t profit until it’s earned. Subscribers cancel, and a first year is a forecast until the year is over.
- One week’s subscription rate is one week. If the rate moves, the target should move with it.
- The bet only works with your real margin and repeat rate in it. That’s why you set the target in the brief yourself.
Valuing orders by margin
If your order events carry the margin on each order, the first-order step can be worth that margin instead of the order value. Agents then weigh every dollar of spend against profit, not revenue, and the target can be set the same way.
Stock and margin also shape the plan as context. The roaster’s brief asks the agent to put “budget behind the products we have the most of and earn the most on.” Its first move that week shifted $60 a day from broad coffee campaigns to the Holiday Blend and gift sets, at the same total spend, because roast-date messages had won first orders for $8 less and gift sets carry the roaster’s best margin.
A return isn’t a profit until you subtract what you paid to earn it.