What disappeared from Meta's privacy settings
There was a control on Meta called "Your activity off Meta technologies." Users who enabled it were disconnecting their off-platform behavior — site visits, product views, purchases — from their Meta profile. For advertisers, those users were invisible in Pixel-based custom audiences. Your retargeting campaigns could not reach them.
Meta removed that setting in July 2026. Its replacement, "Activity from other businesses," controls how behavioral data is used for personalization but does not let users prevent the connection from being made at all.
The rollout excludes the EU, UK, Brazil, South Africa, and Nigeria for now. Everyone else who previously used the opt-out is now back in your targeting pool.
Who's back in your retargeting audiences
Opt-out rates on this setting varied widely by market. In privacy-aware markets, research puts it somewhere between 10–20% of users. That's not a trivial segment.
The users who opted out weren't uniformly low-intent. They were people who navigated to a privacy settings menu, understood what the control did, and turned it on. Some are heavy online shoppers who simply dislike behavioral tracking. Others are high-income, tech-aware buyers who match exactly the demographic many e-commerce brands spend the most to reach.
They're now in your ViewContent, AddToCart, and Purchase event audiences. They also re-enter your lookalike seeds, which means the audience signals feeding your Advantage+ campaigns widened too — quietly and without any notification.
Why your July ROAS might look off
Retargeting campaigns are built around one assumption: the audience is warm. They've been on your site. They've seen your products. They're somewhere in a purchase consideration cycle.
When a batch of previously excluded users enters that audience, the algorithm treats them the same way it treats everyone else in the segment. But these users haven't been retargeted recently. Their in-market status at this moment is unknown. The platform has purchase history data connecting them to your Pixel events, but not recency signals from the past year or two.
What this typically creates is a composition shift, not a collapse. Your retargeting pool is larger. Reach increases. CPMs may drop slightly as pool size expands. But the blended conversion rate on the wider audience can fall below your historical baseline, pulling your reported ROAS down without an obvious trigger.
The honest framing: this is a qualitative tailwind with unquantified magnitude. Most advertisers who held spend stable saw ROAS stabilize within two to three weeks as the algorithm learned which returned users were actually in-market. Cutting retargeting budgets in response to a short-term dip — before that stabilization period ends — is probably the wrong move.
Four things worth checking in your account
Open Meta Audiences and look at size trends for your main retargeting segments over the past 30 days. ViewContent 30-day, AddToCart 30-day, Purchase 180-day. An audience that grew 10% or more without a corresponding site traffic increase is being populated by opted-back-in users.
Check your frequency caps. A larger pool means each individual user sees your ads less often at the same daily budget. If your frequency caps were tuned for a smaller, hotter retargeting audience, they may need revisiting for the new pool size.
If you built any exclusion audiences specifically to remove users who had opted out of personalization, verify whether those exclusions are still catching the right group. The underlying signal those exclusions relied on may no longer exist.
Pull your lookalike performance separately from your WCA retargeting. Lookalike seeds that include opted-back-in users are now more representative, and performance there may actually improve. Mixing both into a single reporting view makes it harder to see what's happening in each.
The free Gromerce audit shows your audience health and ROAS benchmarks together — useful for separating a policy-driven shift from a structural account problem.
Your retargeting pool just got larger. Whether that's a tailwind or a measurement distraction depends on how you read the next few weeks of data.
Sources: Common Thread Collective, AdBeacon, Digital Applied, trackbee, Mintec, TechTimes, July 2026
