On 12 November, Meta switches on a Pixel setting that will make many ad accounts report more purchases. Your sales stay the same, and Meta will see and credit a larger share of them to its ads.
The setting is called “Measure conversions on customer devices”. It sits in Events Manager, among the Pixel settings, and according to the notices advertisers received in October it is on by default. Meta’s help pages, as quoted by Stape and MetricMaven, say it is not active for people located in the EU and the UK.
For a European advertiser that only sells at home, that sounds like someone else’s problem. It becomes yours as soon as part of your spend reaches the United States, Canada or Australia, because your account will then mix two ways of counting from one day to the next. Black Friday falls on 27 November, fifteen days later.
Key takeaways
- From 12 November, Meta counts some purchases on the shopper’s device, by default, outside the EU and UK.
- More reported purchases will not mean more orders. Check your back office before you raise budgets.
- Accounts that mix US and European traffic will see blended ROAS shift for reasons unrelated to performance.
- Annotate the date now and measure a baseline per country.
What Meta is changing
We could not read Meta’s help article directly, so what follows comes from trade publications that quote it, and we say who reported what.
The principle, as described by Stape and Relevant Audience, is that the match between an ad interaction and a later purchase happens on the customer’s device instead of only on Meta’s servers. It relies on information the customer already gave the business, such as an email address, plus temporary device information that Meta describes as non-identifying. Meta then receives the fact that a purchase happened and which campaign gets the credit, without the identifying details.
PPC Land reports that the settings panel shows the line “Measurement starting November 12, 2026”, for datasets that already send events, with no code to add. The attribution window and model do not change, the Pixel and the Conversions API stay in place, and you can switch the setting off pixel by pixel.
The limits deserve more attention than the headline. MetricMaven, citing Meta’s help centre, says the Pixel has to be installed directly on the page, not inside an iframe or as an image pixel. The site also has to require people to log in with an email address or a mobile number.
That login condition is the surprise. A shop that relies on guest checkout may see little change if the rule is applied as written, and the first week of data will tell.
The questions Meta has left open
PPC Land lists what the notice does not say. Do the recovered conversions feed bidding, or only reporting? Is the purchase value sent along? How are consent signals handled? Which browsers and apps are covered?
Bidding is the question we care about most. Relevant Audience reports that Meta’s help text gives two reasons to keep the setting on, seeing more conversions in campaign reporting and improving “campaign performance with a stronger signal”. That wording suggests these conversions will also reach optimisation. If they do, automated campaigns may start leaning towards the audiences where the new counting applies. In an account that serves both the US and Europe, that probably means the US.
The algorithm follows whatever it can measure. That is why we argued that, as Meta automates more of the funnel, trustworthy measurement becomes one of the few levers you keep.
Mark the date before it happens
The first job takes five minutes. Add an annotation for 12 November wherever someone reads Meta figures, in your Looker Studio dashboards, your reporting sheet and your client notes. Whoever opens a report in January and sees ROAS jump on that day should know why without having to ask.
Then measure a baseline. Over the 28 days before 12 November, for each country you advertise in, compare purchases reported by Meta with the orders recorded in your back office or CRM. Not every order comes from Meta, so the ratio will never reach 1. What matters is how it moves.
Take an illustrative case. Before 12 November, Meta reports 60 purchases for every 100 US orders. Afterwards it reports 75, while orders stay flat. Those 15 extra purchases come from the new counting, and your campaigns did nothing to earn them.
Use the attribution window your team works with day to day, and keep it fixed.
Tracking & Reporting
Know what Meta counts and what you actually sold
We reconcile Meta-reported purchases with your back office, country by country, so a reporting change never passes for a performance gain.
Reading an account that spans several countries
Picture a European brand spending a fifth of its Meta budget in the US. From 12 November, the US share of reported purchases rises and the European share stays put. The account’s blended cost per purchase falls, and nobody touched anything.
Split every Meta report by country, or at least into two groups, EU and UK on one side and the rest of the world on the other. A blended figure across both will mislead you for weeks.
Compare a US campaign after 12 November with the back-office ratio you measured, never with its own reported ROAS from October.
Hold off on moving budget from Europe to the US on the strength of reported ROAS in late November. Wait until real orders confirm the gap, even with Black Friday pressing.
Our client reports will carry one extra sentence. From 12 November, Meta counts US purchases with a method that does not apply to European buyers, so the two markets can no longer be compared on reported data alone.
Should you turn it off?
For most accounts, we would leave it on. Yiqi Wu of Aimerce, quoted by PPC Land, said he would not opt out by default and suggested comparing Meta-attributed purchases with Shopify orders while keeping server-side tracking running. We agree. Switching it off keeps your series comparable but may deprive the algorithm of signal your US competitors will be feeding it.
We would think again in two situations. The first is a board report built on Meta-reported ROAS with no check against orders, where the jump would tell a false story in peak season. Fix the reporting, or switch the setting off until you can.
The second is a legal team that has not looked at how this matching sits with your consent setup. Meta says the identifying information stays on the device, but has not explained how consent signals are handled. For US traffic, your counsel should answer that one.
Where the toggle lives. In Events Manager, open Data sources, select your pixel, then Settings. Measure conversions on customer devices appears below the Meta Pixel settings heading, and it is set pixel by pixel.
Your order data is the reference
Meta keeps recovering conversions that browsers and regulators have hidden from it, and this time it moves the matching onto the customer’s phone or laptop. The EU and UK exclusion shows where the legal line sits for now, and it may move.
The platform’s figures remain an estimate, and your orders remain the reference. A per-country reconciliation between the two is easier to build before 12 November than a jump is to explain after it. If you want it running before Black Friday, that is the daily work of our tracking and reporting team.
Sources: Stape - PPC Land, 27 October - PPC Land, 28 October - Relevant Audience - MetricMaven
