Google Signals sunset: what changed on June 15, 2026 (and what to do)

Google Signals no longer contributes cross-device identity or demographic data to GA4 reporting. The practical effects are narrow but real: demographics and interests coverage shrinks, cross-device user counts change, and any audience or report that leaned on Signals-derived data behaves differently. Your event collection, conversions, and revenue reporting are unaffected — this was an identity and enrichment layer, not a collection layer.

What Google Signals actually did

Signals was an opt-in feature that let GA4 use data from users signed in to Google accounts who had enabled Ads Personalisation. When one of those users visited your site on a phone and later a laptop, Google could recognise them as one person and stitch the sessions together.

It supplied three things:

• Cross-device identity, above and beyond your own user ID.
• Demographics and interests — the age, gender, and affinity categories in GA4's audience reports.
• Cross-device remarketing audiences shared into Google Ads.

Everything else in GA4 — event collection, sessions, attribution within a device, ecommerce, key events — worked without Signals and continues to.

What changed, concretely

Demographics and interests reporting thins out. These reports were always based on a subset of users; that subset shrank. Expect low-traffic properties to hit data thresholds and show fewer rows, or none.

Data thresholding gets more visible, not less. GA4 withholds rows that could identify individuals. Ironically, properties that had Signals enabled saw thresholding because of it — some of that suppression eases, while the enriched dimensions themselves become sparser.

Cross-device user counts change. Without Signals-based stitching, the same human on two devices is more likely to be counted as two users. Totals for users and new users can shift, and per-user metrics move with them.

Some audiences shrink or stop refreshing. Any audience defined on demographic or affinity criteria loses reach.

Reporting identity behaves differently. GA4's reporting identity settings blended User-ID, Signals, and device ID. With one input removed, the blended and observed options converge.

What did not change

• Event and conversion collection. Every event still arrives.
• Revenue and ecommerce reporting. Untouched.
• Attribution within a device. Campaign attribution works exactly as before.
• Your own User-ID stitching. If you send a user_id, cross-device identity from your data continues to work — and is now the primary way to get it.
• Google Ads conversion tracking and remarketing via your own tags and lists.

How to tell whether you were affected at all

A surprising number of properties never had Signals enabled, in which case nothing about the sunset touches them. Three quick checks:

1. Did your property show demographics and interests data? If those reports were already empty or permanently thresholded, Signals was either off or your traffic was below the reporting threshold. Either way, nothing changes.
2. Do you have audiences built on age, gender, or affinity criteria? Those are the ones that lose reach. Behavioural audiences are unaffected.
3. Did your user counts step at the changeover? Compare the four weeks either side for users and new users, holding sessions constant. Sessions barely move; users moving by a few percent is the fingerprint of lost cross-device stitching.

If all three come back clean, close the tab — this was not your problem. If the second or third turns something up, the responses below are ordered by value.

What to do about it

1. Implement User-ID if you have logged-in users

This is the single highest-value response. If people sign in, send a stable, non-personally-identifiable identifier as user_id on every event after authentication:

gtag('set', { user_id: 'a7f3c9e21b' });

Use an internal identifier or a hash — never an email address, never anything that identifies a person outside your systems. Keeping PII out of GA4 covers what is and is not acceptable here.

With User-ID in place you get cross-device stitching from your own first-party data, which is more accurate than Signals ever was for your logged-in population, and does not depend on a Google feature staying available.

2. Rebuild demographic audiences on behaviour

Audiences defined on age or affinity should be rebuilt on things you observe directly: pages viewed, categories browsed, purchase history, engagement depth, lifecycle stage. Behavioural audiences generally outperform demographic ones anyway, and they are entirely yours. Building GA4 audiences for remarketing covers the patterns.

3. Re-baseline before you panic about a step change

User counts moving at the changeover is expected, not a bug. Note the date in your reporting, annotate the dashboards, and compare like-for-like periods after it rather than across it. If you skip this, someone will spend a week investigating a "traffic drop" that is a counting-method change.

4. Strengthen first-party data collection generally

Signals ending is one instance of a longer pattern: third-party and platform-supplied identity keeps shrinking, and the data you collect with consent, from your own properties, keeps getting more valuable. Where the third-party cookie actually landed covers the wider picture, and Enhanced Conversions is the advertising-side equivalent of the same shift.

Capability	Before	Now
Cross-device identity	Signals + User-ID	User-ID only
Demographics / interests	Signals-derived	Sparse or unavailable
Cross-device remarketing	Signals audiences	Your own lists, Customer Match
Event collection	Unaffected	Unaffected
Revenue reporting	Unaffected	Unaffected

FAQ

Did Google Signals ending break my GA4 tracking?

No. Signals was an identity and enrichment layer on top of collection. Your events, conversions, and revenue continue to be collected and reported exactly as before.

Why did my GA4 user count change?

Without Signals-based cross-device stitching, one person using two devices is more likely to be counted as two users. This is a change in counting method, not in traffic — annotate the date so future comparisons are not misread.

Can I still get demographic data in GA4?

Coverage is much thinner. For most properties the practical answer is to stop relying on GA4 demographics for decisions and build segments from observed behaviour and your own customer data instead.

What replaces Google Signals for cross-device tracking?

User-ID, populated from your own authentication. It only covers logged-in users, but for those users it is more accurate than Signals was, and it is not dependent on a platform feature you do not control.

Do I need to change my tags or container?

Only if you are adding User-ID, which is a configuration change in your GA4 tag rather than a re-implementation. Nothing needs to be removed because of the sunset itself.

Confirm your GA4 tag is still healthy while you are in there: the free tracking audit checks measurement ID presence, duplicate properties, and consent signals on any URL.