Meta Drops the Off-Platform Data Opt-Out: What Changes

Meta is removing the off-platform data opt-out that has anchored its privacy controls since 2019. Announced on June 9, 2026 via the Meta Newsroom, the change retires the “Your activity off Meta technologies” setting and folds it into an expanded control called “Activity from other businesses” — one that governs how Meta uses the data businesses send it, not whether that data arrives at all.

For advertisers, this is one of the more consequential quiet changes of the year. Under the old system, a user who disconnected their off-platform activity still generated Pixel and Conversions API events — but those events reached Meta as anonymous, unmatchable signals that never joined a retargeting audience. When the opt-out disappears, those previously invisible users can start reappearing in Website Custom Audiences, Lookalike seeds, and conversion-matching pipelines.

This guide covers what Meta actually announced, the usage-versus-collection nuance most coverage glosses over, the three surfaces off-platform data now powers, what it means for retargeting and signal quality, why the July dates keep getting conflated with a separate Meta change, and a practical checklist for ecommerce advertisers. Every factual claim below traces to Meta’s own announcement or clearly attributed trade analysis.

Meta’s newsroom post — titled “Better Personalization and Changes to Controls for Your Activity From Other Businesses” — makes three core statements. First, Meta is streamlining its privacy controls and will no longer offer the “Your activity off Meta technologies” setting, which previously let people disconnect the activity that businesses share with Meta from their account. Second, two overlapping settings are being consolidated: the retired opt-out and a control formerly called “Activity information from ad partners” merge into a single expanded setting, “Activity from other businesses.” Third, the data in question — activity like games played or purchases made on other websites — will now inform Feed recommendations and Meta AI responses in addition to ads.

The setting being retired has history. It launched as “Off-Facebook Activity” in September 2019 and rolled out worldwide by January 2020, according to secondary reporting summarizing Meta’s own feature documentation — meaning the disconnect control survived roughly six years and ten months before this consolidation. Mainstream outlets covered the June announcement as Meta scrapping the feature that curbed web tracking, which is accurate as far as it goes but misses the mechanics that matter to advertisers.

One more scoping note: Meta’s primary announcement names only “the US and a number of other countries” for the first wave. Some trade coverage has reported a more specific initial market list, but that detail is secondary reporting, not Meta-confirmed — and the EU is widely expected to be treated separately given GDPR. If you run campaigns outside the US, plan around the direction of the change rather than a hard date.

Here is the distinction most generalist coverage glosses over: businesses have always been able to send Meta a person’s off-platform activity — via the Pixel, the Conversions API, and offline event uploads. The old opt-out never stopped that flow. What it did was break the linkage: a disconnected user’s purchase and browsing events arrived at Meta’s servers but could not be tied to their individual profile, so they appeared to advertisers’ audience systems as anonymous, unmatchable signals.

The new setting narrows the promise further. Per Meta’s own announcement, it controls whether Meta uses the data to personalize a person’s experience — businesses can continue sending activity data regardless of the person’s choice. Privacy advocates have criticized both that scope and the control’s placement in settings, but as of publication Meta has announced no reversal or delay. The table below maps what off-platform data powered before this change against what it powers after, surface by surface — no source we reviewed lays this out end to end, so we built it.

Read the last row twice, because it is the one that reframes the story. The data pipeline is not changing — what changes is the match rate between the events you already send and the accounts Meta can attach them to. That is why this lands as an advertiser story and not just a privacy story: the same Pixel, the same events, a larger fraction of them resolving to real, targetable profiles.

The “why” is written into the announcement itself. Off-platform activity — Meta’s examples are games you play or purchases you make on other websites — previously fed one surface: ad targeting. Under the new policy it feeds three: ads, Feed content recommendations, and Meta AI responses. Meta’s illustrative example is direct — a recent tent purchase leading to more camping-related Reels. The company frames the whole change around personalization at its stated scale of 3.5 billion people using its apps and services daily, a figure it sources to its Q1 2026 investor exhibit.

Our read: this is signal-pool consolidation. Ads, organic ranking, and the AI assistant increasingly compete for the same behavioral inputs, and maintaining a control that severed one input from one surface no longer fit where Meta is heading. Once Feed and Meta AI draw on commerce signals, the wall between “ad data” and “experience data” stops making sense internally — so the control built around that wall went away. That also explains why the update ships as a consumer-settings consolidation rather than an ad-product launch: nothing new was announced for advertisers, no new Custom Audience type, no new API field. The advertiser impact is entirely downstream of existing tools.

The cleanest way to think about the audience impact is as a visibility change, not a data change. Under the old opt-out, a disconnected user who browsed your product pages and added to cart still fired your Pixel — but Meta could not connect those events to their account, so they never entered your Website Custom Audience. From your side of Ads Manager, they simply did not exist.

As the opt-out disappears, that population becomes matchable again. Common Thread Collective’s analysis — the most thorough advertiser-side treatment we found — expects website-visitor retargeting audiences to grow toward “a more complete picture of who visited your site.” How much larger your pools get depends on how many of your visitors had actually used the disconnect control, and no source quantifies that; any specific percentage would be invented. Directionally, though, bigger pools change tactics: if your retargeting segments have been capped tight for frequency control, or your Custom Audience retargeting filters are tuned to a smaller audience, both deserve a re-look once rollout reaches your markets.

There is a creative angle too. A wider pool means a larger share of people seeing retargeting creative who have never seen it before — which cuts fatigue in the short term but also means your highest-intent segments get diluted with colder re-entrants. Segmenting by recency and engagement depth matters more after this change, not less.

The second-order effect runs through Meta’s machine learning. More conversion events resolving to identifiable accounts means better signal-matching for attribution and better training inputs for Meta’s bidding algorithms — the framing in Common Thread Collective’s analysis is more connected events as better fuel for AI-driven bidding. Campaigns that lean hardest on those systems, like Advantage+ campaigns that already run on pooled signals, are the most direct beneficiaries. Lookalike audiences built from customer lists and purchase events should likewise become more representative as more of your actual customers are identifiable in Meta’s system.

Be careful with the measurement side, though. If match rates improve, some portion of conversions that previously went unattributed can start appearing in reporting — which looks like performance improvement without any change in your media. That is the same class of re-baselining problem raised by Meta’s attribution-counting change earlier this year, and it is why holding a clean pre-rollout baseline matters. If you run multi-platform measurement, our roundup of multi-touch attribution data covers how platform-reported numbers diverge from incrementality.

Projecting forward: expect audience sizes, match quality, and platform-reported conversion counts to drift upward in rolled-out markets over the months after launch, and expect Meta to keep consolidating consumer controls as its ad, content, and AI systems share more infrastructure. Advertisers who snapshot their metrics now will be able to tell platform-driven drift from genuine performance change; those who don’t will be guessing.

A lot of secondary coverage blends “Meta’s July 2026 changes” into one story with one date. There are two distinct announcements, and only one of them has a confirmed calendar day. The location-based ad fees — surcharges tied to six markets — took effect July 1, 2026. The opt-out removal covered in this post has no announced day; Meta’s language is “next month” from June 9. If you have seen “July 1” attached to the opt-out removal, that is the conflation at work.

We covered the fees separately in our guide to Meta’s new location-based ad fees, effective the same month. The two changes share nothing beyond a company and a month: one is a pricing surcharge with a hard date, the other is a privacy-control consolidation with a soft window. Keeping them straight matters when someone on your team asks “what did Meta change on July 1” — the correct answer is fees, not data.

None of this requires new tooling — the whole impact flows through infrastructure you already run. The preparation, adapted from Common Thread Collective’s recommendations plus our own client practice, splits cleanly into before, during, and after the rollout window.

The CAPI item is the one most teams under-invest in. If your server-side setup is partial or your deduplication is loose, this change amplifies the problem — more matchable users means more events that can be attributed twice. Our walkthrough of server-side Conversions API tracking covers the dedup mechanics. And if you would rather have a senior team pressure-test your signal infrastructure and audience architecture before the rollout lands, that is exactly the kind of engagement our paid media service runs — audit first, then restructure only what the data says to restructure.