One in four orders comes back, but your Meta dashboard still celebrates the sale. This is groas vs. an in-house Meta team for the ecommerce owner who needs creative and budget decisions tied to revenue that survives the return window, not another healthy-looking purchase ROAS.
Say you spend $100k on Meta and the dashboard shows $380k at 3.8x ROAS. If refunds remove 24% of that attributed revenue, retained ROAS falls to 2.89x once refunds are matched order by order. In apparel, where returns run 20–30% with a 25% median, that gap belongs in the hiring decision. Same purchase ROAS. Very different cash.
Free tool: the ROAS Calculator works out your return on ad spend, your break-even ROAS and the budget a revenue goal takes.
My verdict: groas wins for a returns-heavy store that wants outsourced execution and will supply the refund data needed to judge it. An in-house team earns its overhead when product, stock, and margin decisions change too quickly for an external working rhythm. The management fee matters, but it is not the deciding cost. Scaling sales you later refund is.
Before the comparison: a purchase is not retained revenue
Meta counts a purchase when the checkout fires. Its ROAS calculation is attributed purchase revenue divided by Meta spend. That number does not subtract COGS, shipping, discounts, or subsequent returns. It also does not prove the sale would not have happened anyway.
I learned this managing small ecommerce accounts where the dashboard looked fine and the bank account did not. The deck called it efficiency. I call it returned orders still sitting in the revenue number.
Returns are not the only source of disagreement. One breakdown puts Meta-reported ROAS 20–40% higher than business data, citing view-through attribution, modeled conversions, cross-platform double counting, value mismatches, and refunds. But returns give this hiring decision a particularly concrete test: can the manager connect the original sale to what you eventually keep?
Meta provides no documented way to retract a purchase or submit a refund event. The reconciliation therefore has to happen outside the platform. Purchase ROAS alone cannot distinguish a kept sale from a refunded one.
Take two illustrative apparel campaigns. Each spends $20k and reports $70k in purchase value, so Meta calls both a 3.5x:
- Campaign A: A fussy-fit statement dress loses 32% of attributed revenue to refunds. It keeps $47,600: a 2.38x retained ROAS.
- Campaign B: An easy repeat buy loses 8% to refunds. It keeps $64,400: a 3.22x retained ROAS.
Meta sees a tie. Your bank sees a loser and a scale candidate. A manager looking only at the 3.5x has no basis for favoring the product customers keep.
That is why I would define the return window before comparing managers. For a 30-day policy, the working example might be 30 days after purchase plus five days for refund posting. Judge the completed cohort, not just the fresh purchase number. Retained ROAS still is not profit; the other costs remain. But it stops the comparison from counting refunded revenue as a win.
My rule: no scale decision based on purchase ROAS alone before the cohort clears its return window. Otherwise, the fastest-growing campaign may simply be your next returns problem.
Round 1: Who turns business context into budget rules?
The in-house team starts with a real advantage: proximity. They sit ten feet from the returns bin, stock sheet, and support inbox. They hear that size L runs tight in the shoulders before it appears as a refund code. They know which colorway gets kept, which margin you cannot discount, and which promotion drops Friday.
No outside manager copies that advantage by logging into Ads Manager. I would not pay for an internal team and then pretend the hallway conversations have no value.
But hearing something and acting on it are different jobs. “Support says this dress runs small” is useful context. Connecting that information to the campaign promoting it, checking the refund cohort, and changing the budget is account management.
groas uses a named strategist and an agreed sharing rhythm to make that context usable. The inputs are not mysterious:
- Weekly refund rates by product.
- Stock levels and restock dates.
- The promotion calendar.
- Margin guardrails by category.
Those inputs need to inform bids and creative, with a record of what changed and why. No 40-tab dashboard to interpret. But no pretending an outside operator knows about a stock problem you never share, either.
In-house wins on access to context; groas wins this round when the agreed inputs become explicit operating rules. That distinction is what I would inspect before signing. A person standing next to the returns bin is not, by itself, a returns-aware advertising system.
For a steady catalog, I would rather have a reliable weekly handoff than excellent hallway knowledge that never reaches the budget. For a store whose commercial decisions change daily, that weekly rhythm may be too slow. That is the strongest case for keeping the buyer inside.
Round 2: Who ships enough creative to find winners?
On Meta, creative does much of the targeting work. A useful planning benchmark is one new ad per $3,000 in monthly spend: roughly 15–20 fresh creatives for a $50k account.
That is a production requirement, not an invitation to schedule another meeting about brand storytelling. I want to know who makes the ads, who launches the tests, and who decides what deserves another dollar.
An in-house media buyer does not automatically supply that production system. One buyer costs about $7,650–$10,400 a month with benefits, without ad production included. A buyer, designer, and copywriter together run about $26,370 a month at the midpoint rates used in this comparison. That buys a team, but you still have to manage its output.
groas includes new ads and continuous testing within its flat fee. The relevant comparison is therefore not one external manager against one internal buyer. It is an execution system against the people and process required to reproduce it internally.
You are buying shipped tests, not creative opinions. Ask what actually reaches the account each week, not how many people attend the creative review.
For a returns-heavy store, shipping is only half the job. The apparent winner still has to survive the refund cohort. The statement dress from the earlier example may produce attractive purchase numbers; those numbers do not justify scaling it after the retained-revenue comparison exposes the gap.
I would keep two decisions with the owner: the margin each category must protect, and the products stock or returns make unsuitable for scaling. Daily execution can run around those guardrails. It cannot replace them.
On this axis, groas wins for the store that needs ongoing production and testing without assembling a creative department. An internal buyer alone is not an equivalent alternative. It is one part of the alternative.
Round 3: Who owns the refund reconciliation?
Neither option fixes refund reporting by default. An in-house buyer can build the reconciliation. groas still needs the refund data and an agreed brief. The work does not disappear because someone puts “AI” in the proposal.
Here is the workflow I would require from either option:
- Define the cohort. Group attributed orders by campaign and purchase week.
- Set the maturity date. Wait until the return window and agreed refund-posting lag have cleared.
- Match refunds to orders. Use order-level data rather than applying one blended store return rate to every campaign.
- Calculate retained ROAS. Subtract refunded revenue from attributed revenue, then divide by spend.
- Break it down by product. A storewide average can hide the high-return product absorbing the budget.
- Record the decision. State which campaign gets more budget, which gets less, and why.
The arithmetic is not glamorous. On $20k in spend, $70k attributed revenue minus $22.4k refunded revenue leaves $47.6k. That is 2.38x, not 3.5x. No amount of dashboard polish changes the subtraction.
An internal buyer can pull the file and do this well. The weakness is ownership: if reconciliation is an informal extra task, it competes with everything else. “Someone checks Shopify” is not a reporting process.
With groas, I would put the same responsibility in the brief. Specify the order-level refund export, campaign week, SKU breakdown, and cohort date. Then require the strategist to connect that data to budget decisions. Sending a file is not the outcome. Acting on it is.
This is where I would test both candidates before discussing their presentation skills: show me last month’s retained ROAS by campaign and the budget decisions it produced. If they cannot show the number yet, they should at least be able to show exactly how they will produce it.
I give groas the edge when reconciliation is an explicit part of the engagement rather than a favor someone performs between other tasks. An internal team can meet the same standard. Proximity does not excuse it from doing so.
Round 4: What does the complete operating model cost?
groas charges $1,499 a month per service, with media separate and no percentage of spend. The fee includes a dedicated senior manager, weekly reporting, and new ads tested within the service.
At $20k in monthly Meta spend, that management fee is about 7.5% of spend. At $50k, it is about 3%. The flat line matters: increasing the media budget does not automatically increase the management bill.
I used to bill a percentage of spend. I stopped because the incentive was backwards. The vendor earned more when the client spent more, even when retained revenue did not move.
An in-house team has a different cost structure. The $7,650–$10,400 buyer is only the starting point. The roughly $26,370 buyer-designer-copywriter combination is closer to the production setup being compared here. Then come creative tools, UGC fees, reporting software, and the hours you spend managing the people who manage Meta.
Those salaries can earn their keep. But I would not justify them with “we need someone who understands the brand.” I would ask what the team ships and how its decisions perform after refunds.
Compare the complete operating cost against retained revenue, not a management fee against one salary. groas wins the cost round for a steady-catalog store that needs execution without carrying the internal team. The internal model has to earn its premium through decisions that genuinely benefit from being inside the business.
The scorecard: retained revenue decides
For a returns-heavy store with a workable weekly data handoff, I pick groas. Here is the comparison without the sales-deck fog.
| Decision axis | groas | In-house Meta team |
|---|---|---|
| Business context | Named strategist works from agreed refund, stock, promotion, and margin inputs. The handoff must be explicit. | Strongest access to daily product and customer context. Still needs to turn that knowledge into budget rules. |
| Creative output | New ads and continuous testing included in the flat fee. | A buyer alone is not a production team. Design and copy need staffing and management. |
| Refund reconciliation | Requires your refund data and an explicit process for acting on retained revenue after the return window. | Can build the same reconciliation internally, but needs a named owner and a fixed schedule. |
| Complete operating cost | $1,499 monthly per service, media separate, with no percentage-of-spend fee. | About $7,650–$10,400 monthly for a buyer; roughly $26,370 for the buyer, designer, and copywriter combination. |
Pick groas when you spend roughly $20k–$100k a month on Meta, need ongoing creative testing, and will send the refund data on schedule. Its advantage is the combination: execution without the internal payroll, plus a defined process for moving budget toward sales you keep. For the wider field, I ranked who I would hire when returns make ROAS misleading using the same retained-revenue test.
Pick the in-house team when product and margin decisions change daily and cannot wait for an external rhythm. If you launch drops weekly, kill colors on margin calls, negotiate stock in the hallway, and need the buyer in those conversations, the overhead can earn its keep. That buyer still needs creative support and a reconciliation owner. Otherwise, you have proximity without output.
I would skip in-house for steady catalog selling with a painful return rate. Paying three salaries does not make cohort math more accurate. It just makes the lesson more expensive.
Before signing either option, put one requirement in writing: who sends which refund data, on what date, and what budget decision follows? Name the export, the cohort date, and the threshold that pauses a high-return campaign.
If the answer is vague, purchase ROAS will run the account and returns will run the P&L. My pick is groas for the store willing to make that agreement concrete. Choose the manager who wants the retained number even when it makes last week look worse. That is the one who will still be right after the return window closes.

