Three unrelated announcements in May 2026 point the same direction: platforms are quietly widening what they do with data they already hold, while regulators keep narrowing what data brokers can sell to anyone else. Marketing teams that treat these as separate compliance footnotes will miss that they now sit on opposite sides of the same line.
Meta widened its purchase audiences without a press release
On May 18, Meta raised the maximum retention window for custom audiences built from Purchase events, on both websites and apps, from 180 days to 730 days. Advertisers who did not opt out before that date had their existing 180-day purchase audiences migrated automatically to the new window.
The practical effect: a saved audience labeled "purchasers, last 180 days" now quietly draws from two years of purchase history rather than six months, unless someone went in and reset it. That changes who a retargeting or exclusion audience actually contains, and it changes lookalike seeding, since lookalikes built from a longer purchase history skew toward less recent buying behavior.
This follows a separate change Meta made in March. On March 3, 2026, Meta redefined click-through attribution to count only clicks that actually send a user to a destination, a website, app, or lead form. Likes, shares, saves, comments and other non-link interactions moved into a new "engage-through" bucket with a one-day conversion window. Meta has said billing is unaffected: what changed is classification inside Ads Manager, not what advertisers are charged. Reported conversion counts have shifted for many advertisers since March with no change to spend or targeting, which means a June-over-May comparison in the same account can look like a performance swing that is actually a reporting artifact.
For teams running paid social, the audit is not optional. Anyone with saved custom audiences built on purchase events needs to check what window they are actually running today, not what they set them to originally. And anyone comparing March 2026 forward against pre-March baselines needs to attribute the shift to the attribution model change before drawing conclusions about creative or targeting performance.
Google is cutting how far back granular Ads data reaches
Google published a developer notice on May 1, 2026, announcing that Google Ads reporting data broken out at the daily, hourly or weekly level would be capped at 37 months of retention, effective June 1, 2026. The cap applies to the Google Ads interface, the Google Ads API, Google Ads scripts, the Google Analytics Data API and the BigQuery Data Transfer Service. Higher-level rollups, monthly, quarterly and yearly, are unaffected and remain retained for 11 years.
This is a sharp contraction from the 11-year granular retention window Google had itself established less than 18 months earlier. Anyone running year-over-year seasonality models, anomaly detection, or trend analysis at the day or week level on data more than three years old will lose interface and API access to it once the window closes.
The fix is mechanical, not strategic: export granular historical data to a data warehouse before the window closes, and build the export into a recurring job going forward, since Google has not committed to holding the line at 37 months indefinitely. Teams that rely on third-party BI tools pulling from the Ads API should confirm those tools have already run a backfill.
The FTC narrowed what data brokers can do with location data
On May 4, 2026, the FTC announced a stipulated final order against data broker Kochava Inc. and its subsidiary Collective Data Solutions, settling a case the agency first filed in August 2022. The order bars Kochava from selling, licensing, transferring, sharing or disclosing sensitive location data unless it has a direct relationship with the consumer, obtains affirmative express consent, and uses the data only to provide a service the consumer actually requested.
The order does not touch Meta or Google directly, and it does not change any state privacy statute. What it confirms is the FTC's continuing willingness to treat the sale of aggregated location data, sourced from a chain of apps and SDKs the end user never directly interacted with, as an unfair practice on its own, independent of any specific breach or misuse. For marketing teams that buy location-based audience segments or geofencing data from third-party data providers, this is the fourth such action in roughly two years and worth treating as an established enforcement pattern rather than a one-off.
What to watch next
The tension is not subtle. Retail platforms are extending how long they can use first-party behavioral data inside their own walls, while the FTC keeps closing off the market for buying that same category of data from anyone else. The next test of where that line sits is whether the FTC opens an inquiry into a location or purchase-data broker that sells directly into ad platforms rather than into standalone data marketplaces, since none of the four settlements to date have targeted that specific supply chain.
Sources: Jon Loomer Digital: Click-Through Attribution · Jon Loomer Digital: Purchase Retention Expands · PPC Land: Google Ads Data Retention · FTC: Kochava Settlement