iOS 27 and Meta Ads: What Changed for Meta Attribution, and What It Means for Profit

iOS 27 shipped on 14 September 2026, and within three weeks an ad-tech reporter was joking that programmatic had been "totally cancelled". If you run Meta ads for a Shopify brand, the reality is calmer and more useful. iOS 27 doesn't block Meta, but it is one of three things changing Meta attribution this year, and only one of those three actually costs you money.

Here's what Apple changed, what Meta changed on its own, and how to tell a reporting problem from a profit problem before you lock your Black Friday budgets.

What did iOS 27 change for Meta ads?

Very little directly. Apple listed no new tracking protections in its iOS 27 announcements. The real change was quiet: Safari, and every iPhone browser built on WebKit, now blocks network requests to "identity resolution" companies, the firms that turn an email address you type into a sign-up form into an ID that follows you across sites. Apple's WebKit team confirmed it on 19 October in a post titled Blocking Identity Resolution Infrastructure.

The domains named in the public WebKit bug report belong to The Trade Desk (including UID2), LiveRamp, ID5, Permutive and Audigent. AdExchanger's sources say the list is fetched remotely and could grow to hundreds of data companies without a new iOS release, and only Apple knows who is on it.

Meta's Pixel and Conversions API aren't on the reported list, and there's no evidence iOS 27 blocks them. Meta reported $63.06B of ad revenue for Q3 2026 on 21 October, up 26%, and its results release and earnings call didn't mention Apple once.

What hurts Meta attribution on iPhones is mostly older, and it keeps compounding:

The reason to pay attention anyway is timing. iOS 27 adoption is still low, with estimates between 2.4% and 16.3% of iPhones, so the effect grows as people update through BFCM, on a platform that carries about 54% of US mobile traffic. And because the block list is remote and unpublished, any tool in your stack can go dark overnight with no release note to explain it.

Meta attribution changed in 2026 too, and not because of Apple

Most of the drop brands have seen in Ads Manager this year came from Meta, not Apple. Two changes matter.

On 12 January, Meta stopped returning the 7-day-view and 28-day-view windows through its Ads Insights API, so any dashboard pulling those windows lost conversions overnight.

On 3 March, Meta redefined click-through attribution to count link clicks only, and replaced engaged-view attribution with a new bucket called engage-through. Engage-through covers likes, comments, shares, saves and video views of 5 seconds or more (down from 10), and it has a 1-day window. The practical effect: someone who saved your ad and bought five days later used to count as a 7-day click conversion. Now they count only if they buy within a day, and a purchase two to seven days after a like or save isn't credited to Meta at all. The default for website purchase campaigns is now 7-day click, 1-day engage-through and 1-day view, as Jon Loomer lays out.

Meta's stated reason is to line its numbers up with Google Analytics and third-party tools that only credit link clicks, and billing doesn't change. But it means your 2026 ROAS is being calculated under different rules from your 2025 ROAS, so a year-on-year comparison isn't like for like. Set a new baseline from mid-March, and report click and engage-through conversions in separate columns.

The part that does cost money: Meta CPMs are up

Attribution changes move the numbers you see. Ad prices move the money you keep. Northbeam data shared by Andrew Jennings on 21 October shows Meta CPMs across its advertisers about 25% higher than a year ago, with the holiday spike arriving in September instead of November. Northbeam's H1 2026 report already had median CAC up 10% year on year, and up 33% for the worst quarter of accounts. Meta's own figure is a 14% rise in average price per ad.

Jennings' advice is to recalculate break-even now, because profit per first order may already be gone at these prices. That's the right instinct, and it's where attribution and profit stop being the same conversation.

Same ROAS drop, two different problems

Take a brand doing $500,000 a month in revenue on $125,000 of ad spend. After cost of goods, shipping, payment fees and returns it keeps 55%, so its contribution margin after ad spend is $150,000 a month. Now its Meta-reported ROAS falls from 2.5 to 2.0. There are two very different reasons that can happen:

If brand A panics and cuts a fifth of its spend, and that $25,000 was really returning 2.5x, it gives up $34,375 of margin to save $25,000 and ends up $9,375 a month worse off. Brand B needs the cut, but it can't tell which spend to cut from ROAS alone.

The tell is in Shopify, not Ads Manager. If revenue and MER hold while Meta ROAS falls, you have a reporting problem. If MER and contribution margin fall with it, you have a profit problem. This is why StoreHero puts contribution margin after ad spend, MER and platform ROAS side by side in one daily view, so the two cases look different on the first screen you open.

What to do before Black Friday

1. Don't cut Meta spend because reported ROAS dropped. Check Shopify orders, MER and contribution margin for the same period first.
2. Check your Meta app pixel's data-sharing mode in Shopify (Settings, then Customer events). Several tracking vendors report that Shopify changed the default to "Optimized" in January, which can pause sharing.
3. Run the Pixel and Conversions API together with deduplication, and capture click IDs on the server when a visitor lands. Our server-side tracking guide covers the setup.
4. Re-run your break-even ROAS and break-even MER with today's CPMs, product by product, before BFCM budgets lock.
5. Track contribution margin after ad spend in dollars every week. A percentage can hold steady while the business makes less money.
6. Book one holdout test for January, when CPMs fall back, so next year's budget rests on what Meta actually adds rather than what it claims.

The bottom line

iOS 27 is a reminder that Apple can switch off parts of your measurement whenever it likes, and Meta can rewrite its own scoreboard just as quickly. Neither of them can change how many orders landed in Shopify or what each one left you after ads. Use attribution to choose between ads, and use profit to decide whether to keep spending.

Key takeaways

• iOS 27 doesn't block Meta's Pixel or Conversions API. Its confirmed change blocks identity-resolution vendors like The Trade Desk and LiveRamp in Safari.
• Apple's block list is remote and unpublished, so tools can stop working without an iOS release.
• Meta's own 2026 changes moved more numbers: view windows left the API on 12 January, and on 3 March engage-through replaced engaged-view with a 1-day window.
• Year-on-year Meta ROAS for 2026 isn't like for like. Set a new baseline from mid-March.
• Meta CPMs are up about 25% year on year (Northbeam), and that is the change that costs profit.
• When ROAS falls, check Shopify revenue, MER and contribution margin in dollars before cutting spend.

Sources

• WebKit: Blocking Identity Resolution Infrastructure (19 Oct 2026)
• WebKit Bugzilla 324771: blocked identity domains in Safari 27
• AdExchanger: Apple plans to block hundreds of programmatic data companies (2 Oct 2026)
• Meta: Third Quarter 2026 Results (21 Oct 2026)
• Meta for Business: Simplifying ad measurement (click and engage-through attribution, 3 Mar 2026)
• Meta for Developers: Ads Insights API metric availability updates (16 Oct 2025)
• Northbeam: H1 2026 Meta CVRs are tanking as CACs skyrocket
• WebKit: Tracking Prevention in WebKit
• StatCounter: US mobile OS market share (Sept 2026)