If you export ROAS from Meta Ads Manager and from GA4 for the same week, the two numbers rarely line up. Meta might show a return of 3.1, GA4 might show 1.9, and your finance report shows something else entirely. This is the single most common attribution headache for paid-social sellers, and it is the tip of a larger ecommerce data reconciliation problem.
The good news: the gap is explainable, and once you understand the mechanics you stop chasing a match that can never happen.
Why Facebook Ads and GA4 compute ROAS differently
ROAS is revenue divided by ad spend. The spend side is usually consistent — both tools read your Meta bill. It is the revenue side that diverges, because each platform decides which sales "belong" to your ads using its own rulebook.
Facebook counts view-through and modeled conversions; GA4 mostly can't
Meta's default attribution window is 7-day click plus 1-day view (Foreplay). That 1-day view means Meta claims credit for a purchase made within a day of someone seeing an ad, even if they never clicked it. GA4 has no concept of a view-through conversion — it only records sessions that actually landed on your site.
On top of that, Meta and GA4 both model conversions they cannot directly observe, such as iOS opt-outs or blocked pixels. Meta tends to model aggressively to fill its window. The result is that platform-reported conversion tools run higher than analytics tools — one analysis found Meta reports roughly a quarter more conversions on average than downstream analytics (Ruler Analytics). More conversions credited to Facebook means a higher Facebook ROAS.
The attribution model itself is different
Shopify and, by default, most sellers think in last-click terms: the last channel touched gets full credit. GA4's default is data-driven attribution (DDA), which splits one sale's credit fractionally across every touchpoint. So a single order might show as 0.4 of a conversion to Paid Social in GA4 while Meta counts the whole thing.
This is the same structural mismatch that makes Facebook ad sessions disagree with GA4 and makes Google Ads conversions diverge from GA4. Different credit rules, same underlying reality.
GA4 loses events that Facebook and Shopify keep
GA4 runs client-side JavaScript, so ad blockers, Safari tracking prevention, cookie-consent declines, and shoppers who close the tab before the thank-you page all cost it data. Field estimates put ad-blocker and consent-affected traffic at roughly ten to twenty-five percent of users (Audiense/Elevar). Every lost purchase event shrinks GA4's revenue numerator — and therefore its ROAS.
Facebook reports on the click date, not the purchase date
Meta files a conversion on the date of the click or view that earned credit, not the day the order happened. A click on Monday that converts Thursday appears in Meta on Monday; GA4 and Shopify both record it Thursday. Comparing single days will never work. Always compare on trailing 7-to-14-day windows.
A worked ROAS example
Say you sell print-on-demand mugs. You spend $1,000 on Meta ads in one week and drive 100 real Shopify orders at $40 subtotal each. Here is how the same week produces three different ROAS figures.
Facebook Ads Manager. It attributes 55 click-through buyers, 15 view-through buyers, and models another 8 it couldn't observe — call it 78 purchases at $40 = $3,120 in reported revenue. It does not subtract 8 later refunds, and it reports much of this on the click date.
- Facebook ROAS = $3,120 ÷ $1,000 = 3.12
GA4. It loses about 20 buyers to blockers, consent declines, and closed tabs, then splits the survivors fractionally under DDA. Paid Social might land around 48 conversions at $40 = $1,920.
- GA4 ROAS = $1,920 ÷ $1,000 = 1.92
Shopify (last-click). It credits Facebook only for buyers whose last click was a Meta ad — the 55 click-through orders at $40 = $2,200, before refunds.
- Shopify last-click ROAS = $2,200 ÷ $1,000 = 2.20
Three numbers — 3.12, 1.92, 2.20 — for one week of 100 orders. The spread between Meta's 3.12 and GA4's 1.92 is not an error. It is view-through, modeling, click-date timing, and DDA all pulling in opposite directions.
What size gap is normal?
A twenty to thirty-five percent gap between Meta-reported purchases and Shopify orders is normal on the default attribution window (Vaizle; TrackBee). For GA4 versus your server-side truth, a fifteen to thirty percent shortfall is healthy, thirty to forty warrants investigation, and forty percent or more signals a real tracking break (BlueFrog; Consentmo).
Translated to ROAS: expect Facebook's ROAS to sit meaningfully above GA4's, with Shopify's last-click view somewhere in between. If Facebook's ROAS is double or more GA4's, check for a Pixel-plus-CAPI deduplication problem before you trust either number.
The dedup trap. If you send the same Purchase from both the browser Pixel and server-side CAPI without a shared event_id, Meta can count it twice. Meta only merges the two copies when they carry a matching event_id and event_name, and only within a 48-hour window (Meta for Developers). A Facebook ROAS that suddenly doubled after you added CAPI is almost always double-counting, not a real lift.
Which number should you actually trust for ROAS?
None of them alone. Each answers a different question:
- Facebook's ROAS answers "how many sales did my ads plausibly influence?" — generous, window-based, optimistic.
- GA4's ROAS answers "how many sales can I directly observe and attribute?" — conservative, lossy, last-touch-ish.
- Shopify answers "how many orders and how much revenue actually happened?" — the server-side source of truth for the sale itself.
The mistake is picking one and calling it your return. The real move is to anchor on Shopify's order count and revenue as ground truth, then treat Meta and GA4 as directional influence signals around it. Better still, layer in the cost side — refunds, processing fees, and product cost — because ROAS on revenue hides whether the ad was actually profitable. That is the same reasoning that applies when Google Ads revenue doesn't match GA4: the platform's revenue figure is not your bank deposit.
From ROAS to true profit
ROAS tells you nothing about margin. A 3.0 ROAS on a product with thin margins after Printify costs, shipping, and Shopify Payments fees can still lose money once refunds hit. This is where reconciling channels by hand breaks down — you would need to stitch Meta's spend, Shopify's orders, and your supplier costs together every week.
PodVector does that stitching for you. It connects Shopify, Meta Ads, Google Ads, Printify, Printful, and Stripe, then computes true per-order profit — revenue minus product cost, fees, and ad spend — so you are not arguing about which ROAS is "right." Victor, its AI operator, reads that live data, flags where a campaign's real return diverges from its reported ROAS, and proposes moves you approve. Victor does not touch your ad account; the actions he executes are Shopify-side, with your sign-off. It is not a dashboard you have to babysit — it is an operator that works your numbers with you.
If you are consolidating storefronts as you clean up attribution — for example moving an Etsy shop onto Shopify — getting one source of profit truth in place first saves you re-doing the reconciliation later.
FAQs
Why is my Facebook ROAS always higher than GA4?
Because Meta credits itself for view-through conversions and modeled conversions that GA4 never sees, and it reports them on the ad-click date. GA4 uses last-click or data-driven attribution and loses events to ad blockers and consent declines. More credited revenue over the same spend means a higher Facebook ROAS. A gap in that direction is expected, not a bug.
How big a difference between Facebook and GA4 ROAS is acceptable?
There is no fixed number, but as a frame: a twenty to thirty-five percent gap between Meta purchases and Shopify orders is normal (Vaizle), and GA4 running fifteen to thirty percent below server-side truth is healthy (BlueFrog). If Facebook's ROAS is more than double GA4's, suspect a deduplication or tracking issue rather than a real result.
Will setting up the Conversions API make my ROAS numbers match?
No. CAPI recovers events lost to blocked browsers, which narrows tracking gaps, but it does nothing about the methodology gaps — view-through, modeling, click-date reporting, and different attribution models. Even flawless tracking leaves a structural gap between Meta and GA4. And if CAPI is set up without proper deduplication, it can make Facebook's ROAS look higher by double-counting.
Should I compare ROAS day by day between the two tools?
No. Meta reports conversions on the click or view date, while GA4 and Shopify record them on the purchase date, so daily figures will never align even when weekly totals converge. Compare on trailing 7-to-14-day windows instead of single days.
Which ROAS number should I use to make budget decisions?
Use Shopify's actual orders and revenue as your anchor for how many sales happened, and treat Meta and GA4 as directional signals of influence. Best of all, decide on profit after costs, not revenue-based ROAS — a healthy ROAS can still be unprofitable once product cost, fees, and refunds are subtracted.