By the Mayab Media Team
On November 12, a toggle most advertisers have never looked at switches on inside Meta Events Manager. It's called "Measure conversions on customer devices." And it's on by default for every dataset that already sends events.
That's more or less the whole announcement. Meta's notice started circulating around October 18, picked up by Wortfilter and the agency ASTIG, and PPC Land pulled the details together on October 27, without so much as a newsroom post from Meta itself.
The quiet rollout is exactly why you should care.
What the toggle actually does
Today, Meta ties a purchase to an ad using signals from your Pixel and, if you set it up, the Conversions API. Both send data to Meta's servers, and the matching happens there. The new setting adds a second route: matching on the shopper's own phone or laptop, using an email address the customer already gave you plus short-lived device data Meta calls non-identifying. By Meta's description, the identifying part stays on the device, and all Meta gets back is the fact of a sale and the campaign it credits.
Attribution windows and models don't change, Meta says. What changes is how many purchases it can connect to an ad, which is the number most small advertisers actually use to judge whether Meta is working.
So expect bigger numbers. Not necessarily more sales.
Your dashboard will look better than your bank account
A quick hypothetical. Say you run an online store spending $4,000 a month on Meta. Last month Ads Manager credited you with 80 purchases, so $50 per purchase. Your store shows 210 total orders from every channel.
After November 12, Meta might report 95 purchases for the same spend. Cost per purchase "drops" to about $42. Nobody bought anything extra. Meta simply linked 15 existing orders to your ads that it couldn't see before. Stape made the same point: reported CPA can fall and ROAS can rise while real sales sit still.
Is that bad? Not by itself. Some of those orders probably did come from your ads, and Meta's own research claims rules-based attribution undervalues ads by a median of 31%. The problem is timing. Black Friday is November 27, about two weeks after the switch. If your numbers jump in that window, good luck separating the holiday, your new creative and a measurement change.
(Everyone loves a pretty ROAS chart in December. Be suspicious of yours.)
The fine print nobody has fully confirmed
MetricMaven lists a few limits, citing Meta's help pages:
- People in the EU and UK are excluded.
- Your Pixel has to be installed on the page itself. Iframe setups and image-only pixels don't count.
- Shoppers have to log in to your site with an email or phone number.
We couldn't load Meta's help article to check these ourselves, and PPC Land noted that none of them shows up in the settings panel text. Treat them as likely, not settled. That last one is a big deal, though. Read literally, it leaves out most small shops that allow guest checkout. If that's you, your reports may barely move.
Meta also hasn't said whether these recovered conversions feed its bidding, whether purchase values come along, or how consent is handled. Real gaps.
What we'd do before November 12
Don't rush to opt out. That's our view, and it lines up with what Yiqi Wu of Aimerce advised on X. Switching it off blindly just means Meta sees less, which could hurt delivery if the data does reach optimization.
Instead, spend fifteen minutes on this:
- In Events Manager, go to Data sources, pick your Pixel, open Settings and find the new toggle. Confirm whether it's on.
- Export a daily baseline for the last 30 days: Meta-attributed purchases next to actual paid orders from Shopify, WooCommerce or whatever runs your store.
- Write down the ratio. If Meta claimed 80 of 210 orders, that's about 38%.
- Keep the Conversions API running. This doesn't replace it.
- Annotate November 12 in GA4 and in your reporting sheet, so nobody credits a measurement change to a creative test.
After the 12th, watch that ratio instead of ROAS. If Meta's share of orders creeps up a few points while total orders hold steady, the toggle is doing its job. If Meta starts claiming purchases that don't square with your total orders, that's when to dig in. At that point, switching it off for a couple of weeks and comparing is a fair test.
One more thing. If you report Meta results to a boss or a board, tell them now. Explaining a jump before it happens is far easier than explaining it in January, when someone asks why the holiday numbers looked so good and the bank balance didn't.
More reported sales aren't more sales, so judge Meta against your store's numbers, not its own.
Want us to check how this affects your business? Book a free audit.
