Meta says one ROAS, Shopify says another. Do we judge the agency on platform ROAS, blended ROAS or MER?
By Dan BartleyUpdated 3 minute read
The short answer
Judge the agency on blended numbers from your store, MER and new-customer revenue, and use Meta's ROAS for decisions inside the account. The two never match because they count differently: Meta credits sales within its click, view and engagement windows, while Shopify credits the last visit it can see. Agree the targets and the attribution setting in writing.
Why do Meta and Shopify never agree on ROAS?
They answer different questions. Meta's standard attribution credits a purchase within 1 or 7 days of a link click, 1 day of a view, or 1 day of a non-link click or five-second video play, while Shopify's marketing reports default to the last non-direct click it can see.
| Reason | Effect |
|---|---|
| View-through and engage-through credit | Meta counts sales from people who saw or interacted with an ad but did not click |
| Visits Shopify cannot connect to an ad | Other devices, blocked tracking and direct visits leave social under-credited in Shopify |
| Different windows and models | The same sale is credited by one, both or neither |
| GST, shipping and refunds | Revenue may include them in one report and not the other, and refunds land later |
Which number should we judge the agency on?
Blended MER and new-customer revenue, both from your store, because they show what the business took for what it spent. Platform ROAS stays useful for comparing ads, ad sets and creative inside Meta, but on its own it lets an account look healthy while the business stands still.
| Measure | How it is worked out | Use it for |
|---|---|---|
| Platform ROAS | Revenue Meta attributes, divided by Meta spend | Comparing ads, ad sets and creative |
| Blended ROAS or MER | Total store revenue, divided by total ad spend | Judging the account and the agency |
| New-customer CAC | Ad spend, divided by new customers | Judging growth rather than repeat sales |
| Contribution after ads | Gross margin, minus ad spend | Knowing whether it all made money |
Split new and returning customers wherever you can. Returning buyers are the sales most likely to happen without an ad, so an account that grows revenue mainly from them can show a strong ROAS while bringing in few new customers.
How do we turn an MER goal into a platform ROAS target?
Work out the MER the business needs from its margin, then scale it by the share of store revenue Meta usually claims. If Meta is your only paid channel, the platform ROAS target is the MER target multiplied by Meta-attributed revenue as a share of total revenue, using the last few months as the ratio.
Recheck the ratio every quarter and after big changes to attribution settings, offers or channels, because it moves. The quick version of this sits under breakeven ROAS and MER targets.
How do we know the ads are causing sales at all?
Test it. Meta now offers an incremental attribution model that optimises and reports on the conversions it predicts an ad caused. A holdout, where ads pause in one region or for a set period, shows what sales look like without them.
Run holdouts outside sale periods and keep everything else steady while they run. A small drop in sales during a holdout means the ads were claiming purchases that would have happened anyway.
What should we agree with the agency up front?
The measures, the targets and the attribution setting, in writing, before the first report.
- The attribution setting for every ad set, and that it will not change without notice
- MER and new-customer CAC targets, worked out from margin
- A monthly reconciliation of Meta's figures against Shopify and actual takings
- Whether every figure is reported with or without GST
- How targets change during sale periods
Other questions people ask about this
Should we switch to 7-day click only?
It gives a stricter view, but the attribution setting also informs delivery, so changing it changes who sees the ads. Keep the setting you optimise on and use Meta's compare attribution settings to read the click-only figure beside it.
Read the full answerIs Google Analytics more accurate than both?
No. It is another model with its own blind spots, mostly the same click-based gaps as Shopify, so use it as a third view rather than a referee.
What is a good MER?
One that leaves profit after product cost, freight and overheads, so it comes from your margin rather than a benchmark. Work out the MER at which you break even and set the target above it.
Read the full answerStill have a question? Ask us