How to use iOS14 ROAS Impact
Apple's ATT prompt cratered the signal that channel-reported ROAS depended on. Here's what actually changed, how to interpret post-iOS14 numbers, and why blended ROAS and MER are now the default lens.
iOS14 ROAS Impact
The breakdown of channel-reported ROAS after Apple's App Tracking Transparency cut the deterministic signal ad platforms used to attribute conversions.
iOS14 ROAS impact refers to the structural break in paid-channel ROAS reporting that followed Apple's April 2021 App Tracking Transparency (ATT) rollout. Once iOS users had to opt in to cross-app tracking — and most didn't — Meta, TikTok, Snap and other platforms lost the deterministic IDFA signal their pixels used to match clicks to purchases.
The consequence wasn't that performance collapsed; it was that visibility collapsed. Reported ROAS in ad managers dropped 30-60% on iOS-heavy accounts even when real sales were stable or growing. The industry response was a shift from channel-reported ROAS to blended ROAS and MER (Marketing Efficiency Ratio) as the source-of-truth metric for online stores.
Before April 2021, the Facebook pixel could match an iPhone user's ad click to a purchase 24 hours or 28 days later using their IDFA — the device-level ID Apple shipped to every iOS device. Attribution was effectively deterministic, and a 4.0 ROAS in Ads Manager broadly matched what Shopify reported.
ATT changed the default. Users now see a prompt asking whether an app can track them across other apps and websites. Industry-wide opt-in rates settled around 20-30%. For the other 70-80% of iOS traffic, the platforms model conversions probabilistically — which is exactly as fuzzy as it sounds.
What actually broke
Three things broke at once, and conflating them is how teams end up chasing the wrong fix. First, the attribution window collapsed. Meta's default dropped from 28-day-click / 1-day-view to 7-day-click / 1-day-view, which alone removes a chunk of credited conversions on longer consideration cycles.
Second, signal volume dropped. With most iOS users opted out, the pixel sees a fraction of the events it used to. Conversions API (CAPI) sends server-side events back to Meta, but without a user identifier those events still need modelling to be matched to an ad impression.
Third — and this is the one most teams miss — the relationship between channel-reported ROAS and incremental revenue weakened. Even if you could perfectly attribute every iOS conversion, you'd still be measuring last-touch credit, not lift. Geo-holdouts and incrementality tests routinely find that channel-reported ROAS overstates true contribution by 30-50% on prospecting and 2-3x on retargeting.
The reporting drop ≠ the performance drop
If your Meta-reported ROAS fell from 4.0 to 2.2 in May 2021, it doesn't mean half your sales evaporated. Most of that gap is missing credit — the same conversions still happen, the pixel just can't see them. Validate against your store's actual revenue before cutting spend.
How to read post-iOS14 numbers
Treat Meta-reported ROAS as a directional signal for creative testing and campaign-level comparison — not as a financial truth. Within a single account, relative changes still tell you which creatives, audiences, and placements are winning. Across the business, you need a different lens.
That lens is blended ROAS (total revenue ÷ total ad spend) or its inverse MER (Marketing Efficiency Ratio). Both ignore platform attribution entirely and ask the only question that matters: for every euro you spent across all channels, how many euros came back through Shopify? It's coarse but it's honest.
Meta-reported vs. blended ROAS — typical DTC apparel account, 2021
Meta-reported ROAS
Blended ROAS (MER)
The chart above is the pattern almost every Shopify store on Meta saw in 2021. Channel-reported ROAS collapsed in April-May; blended ROAS barely moved. The business was fine. The dashboard was the problem.
Why MER became the DTC default
MER is platform-agnostic, so it doesn't care whether Meta's pixel saw the conversion. It's also self-correcting against double-counting: when Meta, Google and TikTok all claim the same purchase (and they do — about 15-25% of conversions show up in more than one ad manager), MER ignores all three claims and just measures total revenue against total spend.
The trade-off is granularity. MER can't tell you whether your TikTok prospecting is pulling its weight against your Meta retargeting. For that you need incrementality testing — geo-holdouts, scaled campaign pauses, or matched-market lift studies layered on top of the MER trend line.
Typical post-iOS14 reporting gaps by vertical (Meta-reported ROAS ÷ blended ROAS share of revenue)
| Vertical | iOS share of traffic | Reported ROAS drop vs. pre-ATT | Under-reporting estimate |
|---|---|---|---|
| Apparel & fashion | 55-65% | 40-55% | 35-45% |
| Beauty & skincare | 60-70% | 45-60% | 40-50% |
| Home & lifestyle | 45-55% | 30-45% | 25-35% |
| Electronics & gadgets | 40-50% | 25-40% | 20-30% |
| Food & beverage | 50-60% | 35-50% | 30-40% |
The pattern is intuitive: the more iOS-heavy the audience, the bigger the reporting gap. Beauty and apparel stores took the worst hit because their customers skew toward iPhone-dominant demographics. If your Shopify analytics show iOS at 60%+ of sessions, assume Meta is under-reporting by at least 35%.
What to do about it
Set up Conversions API alongside the pixel — it won't restore deterministic attribution, but it improves Meta's modelled conversions by 10-20% and helps the algorithm optimise toward real buyers. Shopify's native Meta integration handles CAPI in a few clicks; WooCommerce and Magento have official plugins.
Then move your weekly review from Ads Manager into a blended dashboard: total ad spend, total revenue, MER, new-customer MER (nCAC vs blended new revenue), and a 7-day rolling trend. Channel-level ROAS goes into a separate tab for creative diagnostics, not for budget decisions. Pair this with shorter attribution windows in your day-to-day reporting so you're not comparing apples to a pre-ATT orange.
The honest stack
Blended ROAS / MER as your KPI. Channel ROAS as a creative diagnostic. Quarterly incrementality tests (geo-holdouts or scaled pauses) as your ground-truth check. That's the post-iOS14 measurement stack that actually holds up.
Frequently asked questions
Reported ROAS on iOS-heavy accounts dropped 30-60% between April and June 2021, with apparel and beauty stores at the high end. Real revenue was largely unchanged — the gap is missing attribution credit, not missing sales. Validate against your Shopify revenue before drawing performance conclusions.
Partially. CAPI sends purchase events server-side, which improves Meta's modelled conversions by roughly 10-20% and feeds the optimisation algorithm better signal. It doesn't restore deterministic, user-level attribution — ATT-opted-out users are still anonymous to Meta.
Blended ROAS is total store revenue divided by total ad spend across all channels. Channel ROAS is what each platform reports to itself, which double-counts conversions across Meta, Google and TikTok. Blended ROAS is platform-agnostic and matches your P&L; channel ROAS doesn't.
Yes — for relative comparisons within a single account. It's useful for ranking creatives, testing audiences, and comparing campaign performance over time. It just shouldn't drive budget allocation between channels or be reported to leadership as a financial metric.
Each platform uses last-touch attribution within its own pixel — so if a customer clicks a Meta ad, then a Google ad, then buys, both platforms count the sale. About 15-25% of DTC conversions show up in more than one ad manager. Blended ROAS sidesteps this by ignoring platform claims entirely.
Meta defaults to 7-day-click / 1-day-view, which is the right starting point for most stores. Longer windows over-credit Meta on considered purchases; shorter windows under-credit it. The bigger decision is to stop relying on the window in isolation and pair it with a blended view — see our attribution windows guide for the full breakdown.
MER tells you total efficiency but not which channel is pulling its weight. Incrementality testing — geo-holdouts, scaled pauses, matched-market lift — measures the causal contribution of a specific channel. The honest stack uses MER weekly and incrementality tests quarterly to ground-truth channel allocation.
Less severely. Google has its own logged-in identity (Google account) and ATT only affects in-app tracking on iOS, not browser-based attribution on Safari (which is governed by ITP, a separate restriction). Google Ads reporting dropped 5-15% on most accounts versus 30-60% on Meta.
Divide total revenue for the period by total ad spend across every paid channel for the same period. A 2.5 MER means €2.50 of revenue per €1 of ad spend. Track it weekly; the trend matters more than any single week. New-customer MER (revenue from first-time buyers ÷ spend) is the harder, more useful version.
Yes, for Google Ads and any web-based attribution that still relied on cross-site cookies. The trend is one direction: less deterministic signal, more modelled attribution, more weight on blended metrics and incrementality. Teams that already moved to MER for iOS14 are well-positioned; teams still living in Ads Manager aren't.
Track CAC, channels, and funnel conversion in one place
Metricuno connects ad spend, funnel events, and revenue so you can see CAC by channel, cohort, and campaign — without stitching together five tools.