ROAS, Meta Ads (Facebook)

Metrics type: Key Metrics • Category: Ad Platform

​

At a glance

Return on ad spend, the headline efficiency number for Meta. action_values[purchase] ÷ spend. A 4× ROAS means £1 of Meta spend produced £4 of Meta-attributed purchase value. Below 2× is unprofitable for most DTC margins after COGS, fulfilment, and overhead. Caveat: post-iOS 14.5, Meta’s claimed ROAS over-states by 30, 80% versus commerce-platform truth without Conversions API.

​

Calculation

Calculated automatically from your Meta Ads (Facebook) data. See the At a glance summary above for what the metric tracks and the worked example below for a typical reading.

​

Worked example

A US homeware brand running Meta on a 7d/1d attribution window. The 30-day window is 02 Apr 26 to 01 May 26. Account currency USD. CAPI is live (rolled out 12 Mar 26). Shopify-side total revenue for the same period: $182k. Of that, UTM-tagged Meta-source revenue:$48k. GA4 Paid Social-attributed revenue: $61k.

1. The headline 3.97× is honest by Meta’s measurement, but the business ROAS is closer to 1.2, 1.6×. Real Meta-driven Shopify revenue (a probabilistic blend of UTM truth and modeled lift) is probably $50, 65k, not$154k. Divide that by spend: 1.3, 1.7×. Don’t quote 3.97× to the CFO without the gap caveat.
2. Returning-audience Advantage+ at 6.32× is misleading. Returning customers were going to come back anyway; you’re paying Meta to defend revenue you’d capture for free via email or organic. Same dynamic as Branded Search on Google Ads. The “true acquisition” ROAS is closer to the cold-audience number (3.50×, or 1.4× on the Shopify-truth basis).
3. iOS share is 63% on this account, structurally high. That makes the iOS gap especially material. The CAPI rollout in March helped (the gap was probably 50, 70% before; it’s down to 25, 35% now). Without CAPI, the headline ROAS would read closer to 5×, even more inflated.
4. Modeled conversions are running at 11% of action_values this period. Reasonable, indicates CAPI is working (modeled fill dropped from ~22% pre-CAPI). If you see modeled fill >25% on a CAPI-live account, the implementation has gaps; check event_id deduplication and CAPI event coverage.
5. The 30-day prior window had ROAS 4.21×, slight decline. Spend up 8%, ROAS down 6%. That’s mild scaling pressure, fine for now. If next 30 days shows ROAS down a further 10%+ alongside spend up, hit the brakes; you’re scaling beyond the efficient frontier. Plan a creative refresh in 7, 14 days (frequency on the cold Advantage+ campaign hit 4.7 last week).

Quick sanity tests:

• ROAS up + spend up = healthy scaling, expand budget if pacing allows.
• ROAS flat + spend down = budget cut without channel deterioration. Channel is still healthy.
• ROAS up + spend down = pulled back wisely from low-quality inventory; can scale back when audience refreshes.
• ROAS down + spend up = scaling beyond efficient frontier. Cap budget, refresh creative, audit Advantage+ output.
• ROAS down + spend flat = something changed in attribution, conversions, or competitive landscape. Investigate before cutting. Check Conversions Trend and Clicks vs Conversions first.

​

Sibling cards merchants should reference together

​

Reconciling against the vendor’s own dashboard

Where to look in Meta Ads Manager: Meta Ads Manager → Campaigns → Columns → “Purchase ROAS (return on ad spend)”. This is the column Meta itself surfaces as the headline efficiency reading. Match the date range to this card’s window and the footer total should reconcile to within sub-percent rounding. Other Ads Manager columns that look similar but aren’t:

• Mobile App ROAS: app-purchase ROAS via SDK; separate ecosystem from web. Not in this card.
• Outbound CTR: clicks that left Meta vs clicks within Meta. Not ROAS.
• Conversion Lift / Brand Lift: Meta-run incrementality studies, separate methodology.

Why our number may legitimately differ from Meta: Cross-connector reconciliation: The ROAS reading on this card is Meta’s view of Meta-driven performance. The same purchase will be claimed differently by every platform that has a tag in the path. The honest comparison set:

---

​

Known limitations / merchant FAQs

Why does Meta say I’m at 15× ROAS but Shopify says I’m at 3× ROAS? The single most common Meta question. Three layers stack:

1. Different denominator definitions. Meta’s ROAS uses action_values[purchase] = Meta-claimed revenue including modeled conversions and 7d-click + 1d-view attribution. Shopify’s “Meta-source revenue” uses UTM-tagged orders only. The two are measuring different sets of orders.
2. iOS 14.5 ATT. Even on the same orders, Meta has Pixel + CAPI + modeling; Shopify has only what’s tagged with the right UTM. iOS users without ATT consent strip UTMs at link click; Shopify counts those orders as Direct. Meta still claims them.
3. Last-click vs UTM-truth. Meta credits any purchase that touched a Meta click within 7 days. Shopify only credits orders where the customer arrived via a UTM-tagged Meta link as their last referrer. A user who clicked Meta on Mon, browsed away, returned via Google search Fri, bought, is Meta-attributed in Meta and Google-attributed in Shopify (or Direct, if the Google referrer was lost).

Practical answer: the “true” ROAS sits between the two. Use a weighted blend (40% Meta-claim + 60% Shopify-UTM) as a working figure. Don’t quote 15× to the CFO. Don’t quote 3× either, both are wrong. Use 5, 7× as the directional read for that example. iOS 14.5 ATT impact, the structural answer: ATT (App Tracking Transparency) gave iOS users a system-level prompt to opt out of cross-app tracking. Around 60, 75% of UK / US iOS users opt out. For Meta, this means:

• The Pixel can still fire on opted-in users.
• For opted-out users, Pixel is blocked from receiving advertiser-id (IDFA), so Meta can’t link the conversion back to the user who clicked the ad.
• Meta backfills with statistical modeling (machine-learning estimates) and Conversions API (server-to-server bypass).

Net effect on this card: without CAPI, expect Meta-reported ROAS to over-state real Meta-driven ROAS by 30, 80% on iOS-heavy audiences. With CAPI live, narrows to 10, 30%. Action: roll out CAPI immediately if you haven’t, and treat ROAS readings as directional, not exact. What’s the practical CAPI rollout playbook?

1. Shopify: Install official Meta Sales Channel app → enable Conversions API → choose “Maximum” data sharing. 30-minute setup. Auto-deduplicates Pixel + CAPI via event_id. Easiest.
2. BigCommerce: Meta Pixel app + manual CAPI via Tag Manager + custom event_id. 2, 4 hours.
3. Adobe Commerce / Magento: Meta extension + custom server endpoint. 1, 3 days.
4. Custom / headless: Build a server-side proxy that mirrors Pixel events. 1, 2 weeks.
5. In all cases: deduplicate via event_id (unique per purchase) and matching event timestamps. Test in Meta Events Manager → Test Events tab; if dedup is working, both Pixel and CAPI events show but Meta credits one.
6. Allow 7, 14 days for retraining before judging the ROAS lift; Meta’s models recalibrate over that window.

How do I interpret modeled conversions on Meta? Meta auto-fills attribution gaps with statistically-modeled values. They’re invisible (rolled into action_values) but typically run 8, 18% of total revenue for Pixel-only DTC accounts. Patterns:

• Pre-CAPI: modeled fill ~20, 30%. Inflated.
• Post-CAPI clean: modeled fill ~5, 12%. Healthy.
• Post-CAPI but still high (>20%): implementation gaps, audit event_id deduplication and event coverage.

You cannot opt out. The model can over- or under-state on a given day; trust the rolling 30-day, not the daily. Why is Advantage+ Shopping so opaque? Same trade-off as Google PMax. Meta’s optimiser auto-selects audience, creative, placement, bidding. You see headline ROAS, you don’t see why or where. The pay-off: well-tuned Advantage+ campaigns out-perform manual setups by 20, 40% on cold-prospecting. The cost: limited optimisation control. If you need granular signal, run a manual broad-audience campaign in parallel (smaller budget, manual targeting) for visibility. Meta vs Google Ads attribution differences, who’s more reliable? Both use last-click within their own window, both will independently claim the same purchase if the user touched both channels. Google Ads is structurally tighter because:

• Search has less iOS impact (search intent is desktop and Android-heavier).
• Google Tag Manager + Enhanced Conversions has better consent-mode handling.
• Modeled-conversion share is lower (5, 12% on Google vs 8, 18% on Meta).

Sum of “Meta-claimed + Google-claimed” almost always exceeds total commerce-platform revenue because of double-claim. Use commerce truth (Shopify, BigCommerce, Adobe) as the denominator, not the sum of platform claims. Can I trust the “today” ROAS? Less than the rolling 7-day. Today’s ROAS is built from incomplete data: Pixel/CAPI events are still ingesting, modeled conversions take up to 72h to converge. The 7-day rolling average is the most actionable. Don’t restructure campaigns based on a single day’s ROAS. Multi-account aggregation, how does ROAS work? Meta ad accounts are single-currency by design. Multi-account merchants run a separate ad account per currency or per region. This card is per-account. To roll up across accounts: weight each account’s ROAS by its spend, then sum. A simple average of ROAS across accounts is wrong because it gives equal weight to a $50k account and a$500 account. Why does my Meta say 15× ROAS but my Shopify says 3× ROAS, again? Worth repeating because this is THE recurring boardroom argument. Short answer: Meta is over-claiming due to iOS modeling and 7-day-click attribution, Shopify is under-counting due to UTM-strip on consent rejection. Real answer is in between. Use the weighted-blend approach (40/60 Meta/Shopify or similar). Roll out CAPI to narrow the gap. Ignore anyone who says one is “the truth”, both are partial views. Pixel vs CAPI dual-implementation, do I get double-counted? Only if dedup isn’t configured. With proper event_id matching, Meta server-side dedupes; you get the union (whichever event arrived first or whichever has more data wins, depending on Meta’s logic). Without dedup, you double-count and your ROAS reads ~80% inflated. Test it: Meta Events Manager → Test Events → check dedup status indicator. The official Shopify Sales Channel app handles this automatically; custom implementations need explicit dedup logic.

---

​

Tracked live in Vortex IQ Nerve Centre

ROAS is one of hundreds of KPI pulses Vortex IQ tracks across Meta Ads (Facebook) and 70+ other ecommerce connectors. Nerve Centre runs the detection layer; Vortex Mind investigates the cause when something moves; Ask Viq lets you interrogate any number in plain English. Start for free or book a demo to see this metric running on your own data.