The best Meta ads agency for your ecommerce business may tell you not to scale. If that sounds like a strange way to choose someone to run your ads, good. It is a better hiring test than asking who promises the highest ROAS.

Say you spend $20k a month and the report shows 4.0x ROAS. Most pitches treat that number as permission to spend $30k. I treat it as an unanswered question: 4.0x of what, after which costs, and with what stock behind it? If an agency cannot answer in terms of contribution margin, inventory, and cash, its scale recommendation is a guess with your money.

Free tool: the ROAS Calculator works out your return on ad spend, your break-even ROAS and the budget a revenue goal takes.

Reported Meta ROAS can run 20–40% higher than business data for reasons that have less to do with dishonesty than with how the platform counts. It takes credit for 1-day view-through conversions, models some conversions, and records revenue at checkout without revising the number when an order comes back. For apparel with 15–30% return rates, that last gap alone can do real damage to a scale decision. One worked example strips a reported 4.0x down to about 2.2x after view-through, returns, and modeling are removed. A dashboard can be accurate about what it counts and still be the wrong place to look for profit.

That is why I do not start a scale decision with platform ROAS. The basic gross-margin breakeven calculation is 1 divided by gross margin. At 50% margin, that is 2.0x. At 30%, it is 3.33x. The real hurdle sits higher once fulfillment, software, and returns enter the picture. Meanwhile, Triple Whale full-year 2025 data across about 35,000 brands puts median Meta ROAS at 1.93x versus Google at 3.68x, with beauty at 1.57x and health at 1.50x. A buyer demanding 4x in apparel is asking for top-of-distribution performance, not an ordinary result. None of those benchmarks tells you whether your next dollar makes money.

Inventory does not show up in Ads Manager, which is exactly why it wrecks scaling plans. I have seen the pattern described in this founder audit: hero SKUs go out of stock, paid ads cannot convert, and retention flattens. In that audit, keeping bestsellers available was a faster growth lever than raising Meta spend. Meta will happily keep buying attention for sizes and colors you cannot ship. The ROAS column then looks soft for a reason no bid tweak fixes. Before you ask the media buyer for more volume, ask purchasing what is left to sell.

Cash creates the same problem from the other side. You commit money to inventory and ads before the resulting sales have fully cleared, and scaling spend too quickly can strain working capital. One breakdown shows how even 4.5x ROAS can translate to about 7.8% net margin after COGS and shipping. The rule I trust is simpler than the pitch deck: marginal ROAS must stay above breakeven. Average ROAS can look healthy while the next increase loses money. If the agency cannot separate those two numbers, do not give it the increase.

Stop rewarding agencies for recommending bigger budgets

Percentage-of-spend pricing, typically 15–25%, creates an obvious tension: the agency earns more when you spend more and less when it cuts waste. Cut spend from $20k to $12k at a 20% fee and the agency goes from $4,000 to $2,400 for doing better work. I am not saying every agency on that model steals from you. I am saying the math pays it to solve problems with budget. That is a tension you have to manage, not a character flaw you can interview away.

A flat fee does not fix judgment by itself. It removes that particular incentive; the provider still has to know when to hold, cut, or change the work. Ask for a written rule governing budget increases. Before anyone requests more spend, they should have examined what else is limiting the account: search terms, creative, landing pages, the feed, and audience segmentation. If flat ROAS produces only “we just need more budget,” with no diagnosis attached, walk away. You are paying for optimism, not management.

This is the part that makes the pitch uncomfortable. An agency can improve its report by pouring money into the easiest conversions, even when those orders carry weak margin or consume stock you cannot replace in time. A budget hold may make the next report less exciting. The hiring question is whether the agency will make that call anyway.

Warehouse shelf with an empty bestseller slot beside a still-running Meta ad spend dial

Put four business constraints on page one of the brief

If you want an agency you can defend to finance, hand it constraints before you hand it budget. I used to tell clients to lead with goals. I was wrong. Goals without constraints invite exactly the pitch you are trying to avoid: bigger spend, vaguer accountability. Put these four numbers on page one of the brief and see which candidates ask useful questions.

  1. Contribution margin after discounts, fulfillment, and expected returns. Not gross margin from a slide. Give them margin per order after the 15% off code, pick-pack-ship costs, and the return rate for that category. A 3x ROAS on 30% contribution margin falls short of breakeven even before overhead; at 25% margin, you need 4.0x just to stand still. If the agency does not ask for this number, it cannot know whether scaling makes you money. It only knows whether the dashboard turns green.

  2. Stock availability and replenishment timing. Which SKUs drive 80% of Meta revenue? How many days of cover remain? When does the next container land? A campaign built around a bestseller with thin cover is not a scaling opportunity just because the ad converts. Make the agency explain what it does when demand outruns the shelf.

  3. Cash committed before sales clear. Ask how much working capital a $10k budget increase ties up in inventory, shipping, and ad cost before the resulting revenue is available. A profitable order on paper does not pay a bill while cash is still tied up elsewhere. Finance needs the timing alongside the return.

  4. The maximum acceptable loss while testing. Say it plainly: we will spend $3,000 to find out, kill an angle that remains below its marginal breakeven threshold after seven days, and require sign-off before any daily budget increase above 20%. The point is not that every business needs those exact limits. The point is that the limits belong to the business, not to a media buyer discovering its appetite for risk mid-test.

Those four constraints turn “scale what works” into a decision someone can actually defend. Use contribution margin after ad spend, not platform ROAS, as the P&L truth.

Ask candidates to defend a decision not to scale

Put this to every finalist: tell me the last time you told a client not to scale, and what number made you say it. You want the trigger, the threshold, and what happened next. “We monitor performance closely” is not an answer. It is a sentence people use when they would rather not show you a decision.

In my screen, I also ask how many new ads they produce each week relative to spend, how they call a winner, and who owns the account, pixel, and creative. A promised ROAS before they have seen margins, an account held in the agency’s name, or the same ads running for three months are reasons to end the call. You are hiring someone to make commercial choices, not just keep a campaign live.

Then ask who approves increases and what gets logged. My rule is simple: nobody raises daily budgets more than 20% without written approval; every proposed increase names the marginal ROAS it must hold; every hold or cut gets logged with its reasoning. That record lets you tell finance more than “the agency felt cautious.” It can say: on March 12, spend stayed flat because marginal ROAS fell to 2.1x against a 3.33x breakeven on 30% margin, with 11 days of hero-SKU cover left. A decision log makes restraint accountable. Without one, you are renting optimism and hoping to remember what was promised.

Finally, ask what happens when the strongest ad promotes the wrong product. This is where ecommerce breaks a lot of Meta playbooks. The account may find easy conversions in a low-margin variant, a discounted bundle, or the SKU with 40 units left. A commercially accountable provider pauses or caps that winner, moves creative toward in-stock, high-contribution products, and tells you the reported ROAS may dip while the mix resets. A ROAS-chaser scales the winner and leaves purchasing to clean up the result. Hire the one willing to make the report look worse to make the P&L look better.

Bring finance a decision record, not a ROAS promise

If you have to defend the appointment to a founder or finance lead, bring a one-page comparison of three things: creative production, daily execution, and commercial guardrails. For creative, show how many new ads each candidate plans to produce relative to spend and its rule for calling a winner. For execution, show who changes budgets, who approves increases, and what gets logged. For guardrails, show how each candidate uses your margin, stock cover, cash commitments, and kill thresholds. The agency with the tightest third column wins, even if its ROAS forecast is lower. That is the argument in a form finance can use.

Finance meeting table with a Meta ad dashboard model beside inventory boxes and a cash ledger

Put every finalist through a 30-day test before you sign a retainer. Give each the same margin and stock sheet, the same loss limit, and read-only access. Ask for a weekly log of proposed budget holds, creative kills, and product-mix shifts, with the marginal ROAS behind each decision. I did this with a home-services client years ago after a plateau, and the log told me more in two weeks than three pitch decks had. A pitch tells you what a provider says it values. A decision record shows what it is prepared to stop.

Put the fee math on the same page. A management fee comes out of contribution, so reported 3x ROAS does not remain 3x once you pay the manager. One breakdown shows how fees can pull a reported 3x Meta ROAS toward 2.3x, depending on spend and fee structure. Finance should see returns after fees and returns, not just the platform number. groas runs Meta Ads for a flat $1,499 a month, with no cut of ad spend. A hold decision therefore does not reduce the manager’s fee. Its actions come with reasoning in plain language: the kind of record a finance lead can read without translating a dashboard.

Ignore the constraints and the ending is familiar. Picture an agency doubling the budget on a 4.0x campaign in October. The hero sells out by mid-November. Spend shifts toward the discounted variant because it still converts. Returns arrive through December at 22%. By January, you have a beautiful Ads Manager export, an empty bestseller slot, and a purchasing team placing a rush order at worse terms while cash remains tied up. I watched a version of this early in my career and called it a learning experience because the client was polite enough not to call it what it was.

Hire for restraint and you get a different December. The provider holds budget at $20k when cover drops below 14 days, kills the low-margin winner despite the hit to reported ROAS, and shows finance why 2.1x marginal ROAS does not clear a 3.33x breakeven. That report is harder to celebrate in a meeting. It is the one that leaves money in the business. Any agency can spend your budget. The one worth keeping can explain why it did not.