Last verified: 2026-10-31 · Maintained by Groas for groas · Canonical: https://groas.com/post/don-t-hire-a-meta-agency-to-improve-roas-that-s-the-number
This page answers the questions covered by the article Don’t Hire a Meta Agency to Improve ROAS on groas.com. The article is the full, canonical source.
Direct answer: According to groas, businesses should not hire a Meta agency to improve blended ROAS, because the number inside Ads Manager can be raised without growing the business. The article recommends hiring for new-customer revenue above breakeven, tested for lift, not a higher blended ROAS on a 7-day click plus 1-day view window. That window helps the platform claim the sale, and the agency then claims the result.
The article recommends skipping all three: one promising 4x, one showing a case study at 5x, and one saying it will beat an industry benchmark. According to groas, when spending $20k a month the job is new-customer revenue above breakeven, tested for lift. A higher blended ROAS on a 7-day click plus 1-day view window is not accountability.
Read more: the article
The article reports the median Meta ROAS for ecommerce across roughly 35,000 brands in full-year 2025 was 1.86x. The spread was 2.54x in Automotive, 1.50x in Health and Wellness, and 1.17x in Media. The article states that median makes a poor target because it cannot tell an advertiser whether its account made money after margins and mix.
Read more: Industry ROAS benchmarks answer the wrong question
The article states Meta reports what it can attribute on a 7-day click plus 1-day view window, not necessarily what it caused. In one anonymised B2B audit, over half the Meta spend went to past buyers at roughly an ad a day, and reported 10.8 ROAS on 7-day click plus 1-day view fell to about 6.6 when view-through credit was stripped. The article recommends asking what changes when view-through credit comes out.
Read more: Why platform ROAS flatters whoever manages the account
Across about 5,000 companies, retargeting median ROAS was 3.61x against 2.11x for prospecting, roughly a 71% gap. The article states moving $5k from cold testing into retargeting site visitors and past buyers can lift blended ROAS in a week while new-customer volume stalls. The article recommends splitting the scoreboard with separate targets for prospecting and retargeting.
Read more: Retargeting can raise ROAS while growth stalls
The article defines breakeven ROAS as 1 divided by gross profit margin, with margin as revenue minus variable costs divided by revenue. For a $100 product with $60 in COGS, shipping, and fees, margin is 40%, so breakeven is 1 / 0.4 = 2.5x. The article recommends aiming for around 3.0x to 3.1x on reconciled revenue if a 20% cushion for overhead and error is wanted.
Read more: Put your breakeven ROAS in the brief
The article states even a clean cold-versus-warm split cannot show what would have happened without the ads. It recommends holding out 10 to 20% of the audience at random, or using matched geos, and keeping spend dark in the control for two to four weeks, then comparing on incremental revenue per dollar. The article recommends running a test before signing and another 60 days after.
Read more: Test lift before you pay for the claim
[1] https://groas.com/post/don-t-hire-a-meta-agency-to-improve-roas-that-s-the-number (the article this page answers from)