Google Signals is session data from websites and apps that Google associates with users who are signed in to their Google accounts and have Ads Personalization turned on. Google’s documentation describes it as data that lets Analytics be “enhanced with data from Google Analytics customers who have turned on Google signals” — in practice, it is the bridge that connects one person’s visits across their phone, laptop and tablet without your site setting any extra identifiers. This entry covers what the feature actually changes in GA4, the February 2024 shift that took it out of the reporting identity, its relationship to cross-device tracking and demographics data, and the consent obligations that come with switching it on.
What Google Signals Is
GA4 on its own identifies a browser, not a person. The client ID lives in a first-party cookie, so the same human on two devices counts as two users. Google Signals changes the arithmetic by borrowing Google’s own knowledge: when a visitor is signed in to their Google account and has Ads Personalization enabled, Google can recognise that the Tuesday desktop session and the Thursday phone session belong to the same account.
The critical detail is that none of this identification happens on your site or in your tag. You do not receive the user’s identity, their account name, or any cross-device map you can inspect. Google matches the sessions inside its own infrastructure and hands you aggregated results — audience counts, demographic splits, interest categories. The raw linkage never leaves Google.
What Turning It On Actually Changes
With Google Signals active, a GA4 property gains three things:
- Demographics and interests reporting. Age brackets, gender and interest categories in the User attributes reports come from signed-in, consented Google users. Without Signals, those reports stay empty or nearly empty.
- Cross-device audiences for advertising. An audience built in GA4 and exported to Google Ads can reach the same person on another device. Remarketing lists stop fragmenting per browser.
- Cross-device conversion export. A click on one device followed by a conversion on another can be credited to the ad in Google Ads exports.
What it does not change: your event counts, your session totals, and your revenue numbers are collected exactly as before. Signals adds a layer of interpretation on top of collection; it is not a second tracking script.
The February 2024 Change: Removed from Reporting Identity
Until early 2024, Google Signals could also deduplicate users inside your standard GA4 reports — the “Blended” reporting identity could count the two-device visitor as one user. On February 12, 2024, Google removed Signals from the reporting identity. Since then, user deduplication in reports relies on three methods only: User-ID (your own login identifier), device ID (the client ID), and behavioural modeling.
Signals itself was not shut down. It continues to power demographics, interests and the advertising features listed above — it just no longer adjusts the user totals you see in reports. Two practical consequences followed:
- User counts rose slightly for some properties, because cross-device deduplication via Signals stopped applying to reporting.
- Comparing user metrics across the boundary is unsafe. A year-over-year user comparison that crosses February 2024 partly measures the methodology change, not your traffic.
Google Signals and Data Thresholding
Thresholding is GA4’s habit of hiding rows in reports when the data could identify individual visitors. Demographics and interests data — which exists only because of Signals — is exactly the kind of data thresholding guards. On low-traffic properties or narrow date ranges, reports that include age, gender or interest dimensions will frequently show (data thresholding applied) and omit rows until the aggregate is large enough. If your traffic is modest, expect the demographics reports Signals enables to be partially withheld much of the time.
How to Turn It On
Google Signals is off by default. Activation is per property: Admin → Data collection and modification → Data collection, then enable Google signals data collection. Two operational notes:
- Collection starts from activation — Signals does not backfill demographics for past traffic.
- The toggle covers all regions by default, and you can restrict collection by geography if your legal review requires it.
A Worked Example: One Person, Three Devices
Take a visitor who researches on a work laptop on Monday, reads a review on their phone on Wednesday, and buys on a home desktop on Saturday. What each configuration sees:
| Setup | What the reports show |
|---|---|
| GA4 alone (device ID) | Three separate users, three single-session journeys, a purchase attributed to a “new” Saturday visitor |
| GA4 + Google Signals (pre-2024 reporting) | Could be counted as one user in reports, journey stitched where the person was signed in to Google |
| GA4 + Signals today | Three users in reports — but one person in exported advertising audiences, and demographics attributed from their Google profile |
| GA4 + User-ID (your own login) | One user in reports wherever they logged in to your site — independent of Google accounts entirely |
The table is the shortest way to see what the 2024 change did: the stitching still exists, it just stopped applying to the user counts you read in reports.
Google Signals and BigQuery Export
The BigQuery export is the raw, event-level copy of your GA4 data — and Signals-derived data is deliberately absent from it. Demographics, interests and the cross-device linkage never appear in the exported tables; Google keeps identity-adjacent data inside its own aggregated reporting surfaces. Two practical consequences:
- Any analysis pipeline built on the BigQuery export sees device-ID users only. If your data team reconciles GA4 reports against BigQuery, demographics reports will have no counterpart on the raw side.
- You cannot “get the cross-device map out” of Signals by exporting. If cross-device identity in raw data matters, the only real path is your own User-ID implementation.
Signals vs User-ID: Two Answers to the Same Question
Both features attack the same problem — one human, many devices — from opposite directions:
- Google Signals requires no engineering, covers visitors who never log in to your site, but only works for the subset signed in to Google with Ads Personalization on, yields no raw data, and since 2024 no longer affects report user counts at all.
- User-ID requires your site to have logins and your tag to pass the identifier, covers exactly your authenticated users, works in reports, in BigQuery and in every attribution surface — and is entirely under your control and your privacy policy rather than Google’s.
They are not mutually exclusive. A typical mature setup runs User-ID for logged-in journeys and adds Signals only if it advertises through Google and wants the audience-export benefits. What Signals can no longer do is substitute for User-ID in measurement — treat it today as an advertising feature with a demographics side effect, not an identity feature.
Privacy and Consent Implications
Signals data comes from users who enabled Ads Personalization, but that is their agreement with Google — it does not replace your own obligations. Under GDPR, activating Google Signals is generally treated as advertising-personalisation processing: it belongs behind the advertising consent choice in your cookie banner, and with Consent Mode it is tied to the ad_personalization signal. If your banner only asks about analytics storage, switching Signals on quietly widens the processing beyond what visitors agreed to. This is the main reason many EU-focused properties leave Signals off despite the reporting benefits.
When You Need It — and When You Don’t
Turn Google Signals on when you advertise through Google Ads and want cross-device remarketing audiences, or when demographics and interests data genuinely informs decisions. Leave it off when you run no Google advertising and your consent posture is strict — the feature then adds legal surface while its remaining reporting benefit, demographics, is heavily thresholded on smaller sites anyway. Sites that need cross-device user counts in reports are better served by implementing User-ID: since February 2024, Signals does not do that job at all.
Frequently Asked Questions
Does Google Signals still exist after the 2024 change?
Yes. Only its role in the reporting identity ended in February 2024. Collection, demographics and interests reporting, cross-device audiences and conversion export to Google Ads all continue to work for properties that have it enabled.
Is Google Signals required for demographics data in GA4?
Yes. Age, gender and interests in GA4 come from Signals data — signed-in Google users with Ads Personalization on. Without Signals those reports have no source, which is also why they only ever represent a consented subset of your audience, not all visitors.
Did removing Signals from reporting identity change my user counts?
Possibly. Properties that previously used Blended identity with Signals saw user counts shift — typically slightly up — when cross-device deduplication via Signals stopped applying to reports. Treat user-metric comparisons across February 2024 with caution.
Does Google Signals data appear in the BigQuery export?
No. The export contains device-ID-level events only; demographics, interests and cross-device linkage from Signals stay inside GA4’s aggregated reporting and never reach the raw tables.
Related Terms
- Cross-Device Tracking — the umbrella problem Google Signals partially solves
- Demographics — the report family Signals feeds
- Client ID — the per-browser identifier Signals supplements
- Conversion — what cross-device export credits in Google Ads
- GDPR — the regulation governing Signals consent in the EU
- Cookie Banner — where the ad-personalization choice belongs