Quick Answer: Starting November 12, 2026, Meta is rolling out a Meta Pixel feature that recovers conversions that don't currently appear in campaign reporting because they never make it back to Meta. Those conversions are measured on the customer's device, using information the customer already shared with the business, like an email address, plus temporary, non-identifying device data. Both stay on the device, and Meta only learns that a purchase happened and which campaign should get credit. It applies to datasets already sending events, needs no integration changes, and can be turned off in Events Manager. Expect Meta to report more purchases from mid-November without selling you a single extra order. Record a baseline before November 12, annotate the date, and judge results against Shopify orders.

Key Takeaways

• This is a measurement change, not a sales change. It can recover credit for purchases that already happened. It can't create new ones.
• It turns on automatically from November 12 for datasets already sending events. If you want out, you opt out in Events Manager.
• The timing lands about two weeks before Black Friday, so BFCM comparisons with last year will mix two measurement methods.
• Meta hasn't published how recovered conversions reconcile with Conversions API events, how they appear in reports, or which browsers they cover.
• "On-device" doesn't mean "consent-free." Keep your cookie banner and consent rules exactly as they are.

Meta just told advertisers it will start finding conversions it's been missing. On paper, that's good news. Reported ROAS goes up, cost per purchase goes down, and nothing about your setup has to change.

That's exactly why I'd treat it carefully. The update rolls out on November 12, two weeks before Black Friday. Somebody on your team is going to look at a November report, see Meta claiming more purchases than last year, and move budget toward it. And your store may not have sold one additional order because of it.

I'm not saying the feature is bad. If it fills a real gap in what Meta can see, the reports get more accurate. But more accurate reporting and more sales are two different claims, and the next six weeks are when they're easiest to confuse.

What exactly did Meta announce?

Meta's notice to advertisers, titled "See more of the conversions you're already driving," says:

• Starting November 12, 2026, a new Meta Pixel feature recovers conversions that don't appear in campaign reporting today because they never make it back to Meta to be measured.
• Those conversions are measured on the customer's device, using information customers already share with you, like an email address, together with temporary, non-identifying device data.
• Both stay on the device. Meta only learns that a purchase happened and which campaign should get credit.
• No updates to existing integrations are required.
• It only applies to datasets already sending events, and you can opt out anytime in Events Manager.

That's the whole announcement. Meta hasn't published a technical specification yet.

What hasn't Meta said yet?

A lot, and it matters for how you read your reports:

• Which browsers and devices are covered. Nothing says how it behaves in Safari, in-app browsers, or across devices.
• How it reconciles with Conversions API events. If a purchase is recovered on the device and the same purchase later arrives from your server, it's not documented how Meta avoids counting it twice.
• Whether recovered conversions are labeled. It isn't clear whether you'll be able to see recovered conversions separately in Ads Manager.
• Whether they feed bidding. The notice talks about reporting. It doesn't say whether recovered conversions are used for optimization or audiences.
• How it treats consent. It doesn't say what happens when a visitor declines cookies or withdraws consent.

Until Meta answers those, the safe reading is narrow: Meta can now credit some purchases it previously couldn't. How many, which ones, and how they show up are still open questions.

How is this different from what you already have?

One point is worth separating out. Today, Advanced Matching and the Conversions API send hashed customer data to Meta. The new feature is described as keeping the customer's email and device data on the device. Those are different data paths. The new one doesn't change how your existing Pixel and server events work.

What will change in your reports?

Here's a simple example of why a better-looking report isn't proof of better performance. These numbers are made up to show the math, not a forecast.

Meta now reports 25% more purchases at a 20% lower cost. Your store sold exactly the same number of orders. That might mean Meta was undercounting before, which is useful to know. It doesn't mean your ads started working better.

It also helps to know who's being recovered. If recovered conversions skew toward customers who already buy from you, Meta can look more effective at acquisition than it is. That's why new versus returning customer reporting matters more after November 12, not less.

Who will see the biggest change?

My expectation, and it's only that until Meta publishes more detail: brands that rely on the browser Pixel alone have the most purchases going missing today, so they're likely to see the biggest jump. Brands that already send every purchase to Meta server-side, with strong customer data for matching, have less left to recover.

Either way, the right question isn't "did Meta's number go up?" It's "does Meta's number now match reality more closely?" Your Shopify orders are the only way to answer it.

Should you keep it on or opt out?

There's no universal answer, so here's how I'd decide.

Keep it on if you have a clean baseline, you're measuring new versus returning customers, your consent setup is solid, and someone will compare Meta's numbers with Shopify orders every week through BFCM.

Consider opting out until January if you're running holdout or incrementality tests through BFCM, you're comparing year over year and can't annotate the change, or your legal team hasn't reviewed the new processing.

If you opt out, do it deliberately and write down when. If you keep it, treat November 12 as the start of a new measurement period.

What to do before November 12

1. Save a baseline. Export daily Shopify orders, net revenue, Meta spend, Meta-reported purchases, and purchase value for at least the last eight weeks.
2. Annotate November 12 in every report, dashboard, and BFCM recap that includes Meta.
3. Check your Events Manager settings directly. Log in the normal way rather than through a link in an email, and look for the new setting on each dataset you use.
4. Make sure your purchase event is clean. Recovery won't fix a purchase that fires on the wrong page, twice, or for orders that shouldn't count.
5. Keep a shared event ID between your Pixel and server events, so Meta can deduplicate the purchases it already receives.
6. Turn on new versus returning customer reporting, so you can see whether recovered conversions are new customers or existing ones.
7. Leave your consent setup alone. Don't loosen your cookie banner to help recovery work. On-device processing isn't automatically exempt from privacy rules.
8. Compare weekly against Shopify. If Meta's purchases climb while Shopify orders don't, that's a reporting change, and your budget should follow Shopify.

Recovery can change how Meta credits purchases. It can't change how many orders you actually got. Aimerce sends every Shopify purchase to Meta server-side from Shopify's order data, with a shared event ID for deduplication, so the purchases Meta already receives stay complete and counted once.

It also gives you two ways to check Meta's new numbers. Aimerce's new and returning customer purchases are based on your Shopify customer list, so you can see whether Meta's extra conversions are new customers. And Aimerce's Attribution page shows Meta next to Google and other channels, built from the same Shopify orders.

FAQ

What is Meta's on-device conversion recovery? A Meta Pixel feature rolling out from November 12, 2026 that recovers conversions missing from campaign reporting by measuring them on the customer's device. It uses information the customer shared with the business, like an email address, and temporary, non-identifying device data, which stay on the device. Meta only learns that a purchase happened and which campaign gets credit.

Do I need to change my Pixel or Conversions API setup? No. Meta says no updates to existing integrations are required. It applies to datasets that are already sending events.

Can I opt out of Meta's on-device conversion recovery? Yes. Meta says you can opt out anytime in Events Manager.

Will my Meta conversions go up after November 12? They may. Recovering previously unreported conversions can raise reported purchases and lower reported cost per purchase without any change in actual sales. Compare against your Shopify orders.

Does on-device measurement mean I don't need consent? No. Meta hasn't said how the feature handles declined or withdrawn consent, and on-device processing isn't automatically exempt from privacy rules. Keep your consent setup as it is.

Should I still use the Conversions API? Yes. Nothing in Meta's notice suggests replacing it. Server-side events remain the reliable way to deliver purchases you can confirm, with a shared event ID for deduplication.

Sources

[1] Astig, "Meta Pixel is getting a feature to catch ad conversions you didn't know you made,"

[2] Metricfixer, "Meta Pixel Conversion Recovery: How It Works, Consent and Safari,"

[3] Meta for Developers, "Handling Duplicate Pixel and Conversions API Events,"

[4] Aimerce, "How To Track New Vs. Return Customers' ROAS on Meta Ads Manager,"

[5] Aimerce, "How to Fix the Shopify vs. Meta Revenue Discrepancy,"

Related reading

• How to Fix the Shopify vs. Meta Revenue Discrepancy
• How To Track New Vs. Return Customers' ROAS on Meta Ads Manager
• Google Demand Gen View-Through Conversions Just Changed
• Consent Mode v2 on Shopify: What Actually Changes for Your Conversion Data
• Meta Customer Lifecycle Strategy Needs Clean Customer Data