How Google signals works
Google signals is a switch inside a GA4 property. Turn it on and Google is allowed to associate the visits it records with people who were signed in to a Google account at the time and who have left Ads Personalisation switched on in their own Google settings. Everyone else is unaffected. Their visits are still collected; they simply carry no signed-in identity.
What that association buys you is recognition across devices. Somebody who reads a page on a phone at lunchtime and enquires from a laptop that evening can be counted as one person rather than two. It also unlocks the age, gender and interest reports, which are inferred from the same signed-in profiles, and it lets audiences built in GA4 be shared with Google Ads for remarketing.
The trade for that is data thresholding. Where a report could expose an individual because the group behind a row is small, GA4 withholds the row instead of showing it. Google has also changed over time how far signals feeds ordinary reporting rather than only advertising features, so check the current documentation before you build anything that depends on it.
Why Google signals matters
For a small business the honest answer is that it matters most for advertising and least for reporting. If you send GA4 audiences into Google Ads, or you want interest data to sanity-check who is actually reading your pages, signals earns its place. If your only use of GA4 is enquiries by channel, you can leave it off and lose nothing you were using.
It matters in the other direction too. A property with signals on and a modest amount of traffic — a Kathmandu service business, a single-city clinic, a site launched last month — can find that its most useful breakdowns simply vanish. Nothing is broken. The numbers are being withheld on purpose, and knowing that saves an afternoon of debugging.
Common mistakes with Google signals
The most common is switching it on and then treating the resulting user count as a headcount. Cross-device stitching only covers signed-in visitors who allow personalisation, so it removes some duplication and not all of it. The count moves closer to reality without becoming reality.
The second is enabling it quietly. Demographic inference and audience sharing are exactly the kind of processing a privacy policy is supposed to describe, and turning them on without a word leaves your published policy wrong. The third is hitting thresholded reports afterwards and blaming the tracking setup for a privacy rule.
How to act on it
Decide from the use, not from the default. If you run Google Ads remarketing, turn it on, record it in your privacy policy and expect thresholding on small segments. If you do not advertise, leave it off and keep fuller reports.
Either way, check which reporting identity the property uses, because that setting and this one interact. And if the demographic reports were the reason you enabled it, treat them as a rough profile of your signed-in visitors rather than of your customers. A careful GA4 setup writes those limits down where whoever reads the report will see them.
