Your Meta ROAS falls from 4.0x to 3.4x, so you cut budget until the dashboard looks healthy again. You may have just cut profitable sales.
Say you run an ecommerce store spending $10,000 a month on Meta. Last month, Ads Manager showed $40,000 in attributed revenue. This month, it shows $34,000 on the same spend. Something changed. But that average does not tell you which spending to cut, or whether cutting is the right move.
I used to give clients that exact advice: get the ROAS back up. I was wrong. Average ROAS describes the whole account; the budget decision depends on what additional spending earns after variable costs. A Meta management company should explain that distinction before selling you aggressive scaling or a prettier number.
Free tool: the ROAS Calculator works out your return on ad spend, your break-even ROAS and the budget a revenue goal takes.
Here is the model I use, one layer at a time.
First, work out what an order leaves behind
Meta reports revenue. Your business keeps what remains after costs. Before judging ROAS, subtract the variable costs attached to an order:
- Product cost.
- Inbound and outbound fulfillment.
- Payment and platform fees.
- A reserve for returns and refunds.
What remains is contribution before ads. That is the pool that pays for advertising, then contributes toward overhead and profit.
Take a $60 average order. Say it carries $18 in product cost, $7 to pick, pack, and ship, and $3 in fees and refund reserve. That leaves $32, or about 53% of revenue, before spending a dollar on Meta.
Divide revenue by that contribution: $60 ÷ $32 gives an order-level break-even ROAS of about 1.88x. The general formula is:
Break-even ROAS = 1 ÷ contribution margin
Use the margin as a decimal: 0.50 for 50%, for example. This is the same contribution-margin calculation, just expressed as an advertising threshold.
Below that return, revenue does not cover the order’s variable costs plus advertising. Above it, something remains for overhead and profit. Clearing this line does not mean the entire business is profitable.

That break-even may sit somewhere quite different from your preferred dashboard target:
| Contribution margin before ads | Break-even ROAS |
|---|---|
| 50% | 2.0x |
| 45% | About 2.22x |
| 40% | 2.5x |
If your margin drops below 33%, even a clean 3.0x loses money before overhead.
I keep management fees separate from this order-level calculation, then include them when judging the whole program’s profit. First ask whether the ad spending earns contribution. Then ask what remains after management fees and other costs. Neither question replaces the other.
Write down your contribution per $100 of revenue before arguing about a ROAS target. Without it, the number is decoration.
Next, separate the whole account from the next slice of spending
Average ROAS blends all spending together. The budget call turns on marginal ROAS: extra revenue divided by extra spend.
For the next example, I will use a hypothetical 50% contribution margin to keep the arithmetic clean. Start with the same $10,000 spend and $40,000 revenue:
- Revenue leaves $20,000 in contribution before ads.
- Subtract $10,000 in advertising.
- You keep $10,000 before fixed costs and management fees.
Now add $2,000 in spending. Suppose it produces $6,000 in additional revenue. That slice runs at 3.0x ROAS. The account average falls from 4.0x to about 3.83x: $46,000 divided by $12,000.
The dashboard looks worse. The contribution math looks better:
| Additional spending calculation | Amount |
|---|---|
| Additional revenue | $6,000 |
| Contribution at 50% | $3,000 |
| Additional ad spend | −$2,000 |
| Additional contribution after ads | $1,000 |
A lower average can leave you with more money. The extra spending earns less efficiently than the original spending, but it still earns enough.
Change the revenue and the verdict changes. If that same $2,000 produces $5,000, it leaves $2,500 before ads and $500 afterward. Still positive, just less attractive. If it produces $4,000, the $2,000 contribution only covers the added advertising. Nothing remains from that slice for fixed costs or fees.
The average can look respectable in all three cases. It cannot tell you which one you bought.
Before changing budget, price the slice you intend to add or remove. Do not ask the whole-account average to answer a different question.
Then check whether Meta caused the revenue it counted
There is a catch in that neat example: “produces $6,000” is doing important work.
Attributed purchases are not automatically incremental purchases. Meta can credit a sale after an ad interaction without proving the sale would disappear without that ad. Cross-channel double-counting, view-through credit, and retargeting can all widen the gap between reported and incremental return.
I treat the dashboard as a credit log, not a profit log. It tells me which sales the platform claims. It does not tell me which dollars would have arrived anyway.
Marginal return only settles the budget question if the additional revenue is genuinely additional. A before-and-after revenue change is not automatically the effect of a budget change.

Attribution settings also change what gets counted. A 7-day click and 1-day view window, for example, credits purchases within seven days of a click or one day of a view. Change the window and reported ROAS can change without real sales changing at all.
Returning customers, promotional periods, and sales reported before fulfillment settles complicate the comparison further. A preorder on a dashboard is not the same thing as a fulfilled order with its final costs known.
Before trusting the apparent return on a budget change, I want:
- The same attribution setting across the comparison.
- Comparable reporting periods, with promotions flagged.
- A change log showing what else moved.
- Returns and fulfillment costs reflected in the margin.
- A holdout or lift test where feasible, with its uncertainty acknowledged.
I do not apply a universal discount to every account’s reported ROAS. The useful question is what the evidence says about this spending, not which haircut makes the spreadsheet look scientific.
Keep the measurement consistent before treating a change in reported revenue as a reason to move money.
Lower ROAS is acceptable when the extra contribution is worth carrying
Return to the hypothetical 50% margin. Break-even is 2.0x. The next $2,000 buys $6,000 in genuinely additional revenue, leaving $1,000 after the added advertising.
On that math, the drop from 4.0x average ROAS to 3.83x is not a reason to cut. It is the consequence of adding a less efficient but still profitable slice.
But positive contribution is permission to consider scaling, not an instruction to scale. The business still needs to fund inventory, ship orders, and absorb returns. A profitable marginal calculation does not make the cash requirement disappear.
Before approving the increase, I want stock, fulfillment, and cash checked alongside the return. Buying more orders than the business can carry is not a clever interpretation of ROAS.
The reverse matters just as much. At a 40% contribution margin, break-even is 2.5x. If the next slice buys at 1.5x, each additional $1 of advertising produces $1.50 in revenue and only $0.60 in contribution. That is a $0.40 loss per added advertising dollar, before management fees and fixed costs.
The blended account might still show 3.5x because earlier spending performs well. That does not rescue the losing slice.
This is why margin-based break-even benchmarks are more useful than a generic industry target. A 3.0x return clears the line at 50% margin and misses it at 30%.
My rule is simple: the average tells me what happened, the contribution margin sets the hurdle, and the marginal return tells me whether the change clears it.
Make a management company explain its next budget move
Any firm pitching better return on ad spend should survive three questions before touching the budget.
- What contribution margin sets my break-even? I want the product, fulfillment, fee, and return assumptions behind it. A firm quoting a 4.0x target without asking about costs is pricing your profit on someone else’s benchmark.
- What comparison supports this recommendation? I want the attribution setting, period length, promotions, and other account changes identified. “ROAS went down” is an observation, not a diagnosis.
- What marginal threshold triggers the next move? I want a stated rule for increasing, holding, or reversing spending, including how much clean evidence the team needs before acting.
An answer might be: “At a 45% contribution margin, we need incremental ROAS above roughly 2.22x. We assess the increase over two comparable weeks before deciding whether to continue.” That is a hypothetical operating rule, not a universal two-week prescription.
No threshold, no justified budget change.
Then I separate the creative call from the spending call. A falling average can reflect fatigue, calling for new angles and offers. It can also reflect spending beyond the point where the next slice pays, calling for a pullback. Those are different diagnoses.
The highest-ROAS ad can be the wrong one to scale because its result may come from reaching the easiest buyers first. I want evidence that separates the explanations: a holdout, a geo split, or a creative test that holds spending flat while changing the asset.
The emphasis on guardrails and logged reasoning is what I want from groas, and what I would demand from any management company. Write down the margin, the measurement window, and the decision rule before money moves. Then record why it moved.
A vendor who can only promise a prettier average has not explained how to manage the budget.

Put the model through a new case
Now change the numbers. Say the account spends $20,000 this month and Meta reports $60,000 in revenue. Average ROAS is 3.0x, down from 3.6x the previous month. This month includes a $5,000 spending increase you want to assess.
The contribution margin after product, fulfillment, fees, and return reserve is an illustrative 45%. Break-even is about 2.22x.
On reported revenue, the whole account clears that line. It can contribute toward fixed costs and management fees. That still tells me nothing about whether the added $5,000 earned its keep.
Suppose the comparison attributes $11,000 in additional revenue to that increase. Reported marginal ROAS is $11,000 ÷ $5,000, or 2.2x. It already falls just below break-even.
Now suppose a holdout estimates that 60% of that attributed increase is incremental. This is an assumption for the example, not a discount to copy into your account.
Work through the same model:
- Incremental revenue: $11,000 × 60% = $6,600.
- Contribution before ads: $6,600 × 45% = $2,970.
- Contribution after added ads: $2,970 − $5,000 = −$2,030.
The average still reads 3.0x. The added slice loses money.
I would reverse that increase, keep the base in place while assessing its own economics, and test the creative explanation rather than assuming broader reach will fix the problem. The evidence here argues against the increase, not automatically against every dollar in the account.
That is the discipline I want before any management company touches budget: state the margin, state the measurement window, state the marginal line, and name who approves the next increase. Include a stock and cash check. Record the reasoning so the next review is not an argument about what everyone remembers.
A vendor arguing from contribution will sometimes tell you to keep spending while ROAS falls, and sometimes to cut while it still looks good. Hire that one, not the one selling a prettier average.
If you run this on your numbers this week, do it on paper before opening Ads Manager. Write your contribution per $100 of revenue, the break-even that follows, and the marginal threshold you will defend over a clean comparison period. Then price only the change, using consistent attribution and flagging promotions. The decision gets boring fast: keep the spending that clears the line, cut the slice that misses it, and test the diagnosis before reaching for more budget.

