Fair Warning
On Thursday Meta starts counting sales on your customers' phones
This Thursday, 12 November, a Meta pixel setting called "Measure conversions on customer devices" is due to start working, and the reporting on it is consistent on the part that matters: it's on by default. The toggle is already sitting in Events Manager for a lot of accounts with a label saying measurement starts on 12 November, and advertisers outside the EU and the UK are the ones in it. PPC Land wrote it up first and a run of ad tech and agency write-ups have followed, several of them quoting Meta's own in-product help text, so I'd treat the date and the default as reliable. What I can't point you at is a Meta newsroom post or a help centre article, because I haven't found one, which is becoming a pattern with the changes that arrive inside the product rather than with an announcement.
Here's what it actually does. At the moment your pixel sits on your website and tells Meta when somebody buys, and a decent share of those purchases never get matched back to the ad that caused them, because browsers block things and people switch devices and cookies expire. This setting does some of that matching on the customer's own phone instead, using information the customer already gave you, an email address at checkout for instance, plus some temporary device data that isn't meant to identify them, and what travels back to Meta is reportedly just the confirmation that a purchase happened and which campaign should get the credit. It runs alongside the pixel and the Conversions API rather than replacing either.
Nothing about your business changes on Thursday, only what Meta can see of it, and the number on the screen is what most people react to.
So the thing to brace for is this: your reported purchases and leads will probably go up, and your actual sales won't. Meta isn't creating conversions, it's catching ones it was already missing, which is genuinely a better picture of reality. But if you look at your dashboard on the 20th and feel like something clicked into gear, it may just be that Meta can now see what was always happening. That's an easy thing to accidentally make a decision on.
The timing is the part I'd actually grumble about. The 12th is about a fortnight before Black Friday on 27 November, so it lands right in the middle of the only three weeks of the year where a lot of small businesses are watching their ad reporting daily and comparing it against last year. Any year on year comparison you make this season will have one measurement method on one side and a different one on the other, and nobody has published what the gap looks like, so there's no number I can give you to adjust for it.
On whether to turn it off, the people who do this for a living are split, and I'd rather tell you that than pretend there's a consensus. Some are saying switch it off until your own lawyer or whoever handles your privacy policy has looked at it, since it's a new way of handling customer data on a customer's device. Yiqi Wu at Aimerce was quoted saying the opposite, that he wouldn't opt out by default and would only do it if the reported numbers stopped matching the real orders. I lean toward leaving it on and keeping a record, because a more complete picture is usually worth having and because the pixel was undercounting you before, but I'd want to know the date it changed so I don't get fooled by my own dashboard later.
And if you don't have a website with a pixel on it, which is a lot of local service businesses, none of this touches you and you can skip the rest of this one.
Do this week
Before Thursday, write two numbers down somewhere you'll find them in January: what Meta is currently reporting as purchases or leads for the last seven days, and what your till, your inbox or your booking system actually recorded over the same seven days. That's your before. Then go to Events Manager, open Data sources, click your pixel and open Settings, and decide whether to leave the toggle on or switch it off. It's per pixel and you can change it back. Either way, note the date in with your records, because the only real risk here is comparing December to November as though they were measured the same way.