GA4 Audiences and Sharing Them with Google Ads
A GA4 audience is a list of visitors who did something you care about: viewed a service page, started checkout, bought twice. Inside Analytics it gives you a report. Shared with Google Ads it becomes a remarketing list you can bid on. This guide covers how to build audiences in GA4, how membership and timing work, what predictive audiences need, how audience triggers work, and what has to be in place before your lists fill up in Google Ads.
What a GA4 audience is (and is not)
Google's introduction to audiences defines one as "a group of users from your site and/or app who have generated similar behavioral data or who share demographic or other descriptive data". You build it from conditions, and GA4 adds users who match them and removes users who stop matching.
An audience is easy to confuse with a segment in an exploration, because both use conditions. They behave differently. Google puts it this way: a segment "doesn't include any membership logic and it's fully retroactive", while an audience "is not calculated retroactively from the date it was created". If you only want to look back at last quarter's buyers, use a segment in an exploration. If you want a list that keeps updating and can be used in ads, create an audience.
Every property starts with two predefined audiences: All users and Purchasers. Google says the Purchasers audience "is shared with any linked advertising products".
Audiences only fill from the day you create them
This is the point most people miss. Google's create, edit and archive audiences page says: "After you create an audience, it accumulates users who meet the conditions from that point onward." So build the audiences you are likely to need before you need them, not on the day you launch a campaign.
There is one exception, and it only covers the copy that goes to Google Ads. Google says that when you create an audience, "Analytics adds any users who have met the audience criteria during the last 30 days. This applies to Google Ads audiences only." The audience report inside Analytics does not show these backfilled users. Google's page on audience size differences says a user "will need to log any event (become active) after the audience is created" to appear there.
Timing to expect: "it can take 24-48 hours for the audience to accumulate new users", and the same delay applies when a new user starts matching. The Summary card in the builder shows how many users met your conditions in the last 30 days, which gives you a rough size before you save.
Creating an audience
You need the Marketer role or above on the property. Then:
- In Admin, under Data display, click Audiences, then New audience.
- Choose a custom audience, a template (for example Demographics) or a suggested audience. Suggested audiences are based on the industry category you set for the property and on your app store categories.
- Add conditions from dimensions, metrics and events, and set the scope: across all sessions, within the same session, or within the same event. Google allows "a maximum of 10 conditions".
- Add a sequence if the order matters (for example viewed pricing, then submitted the contact form), and optionally a time limit for the whole sequence.
- Add a group to exclude, temporarily or permanently. Google says users "are first evaluated by INCLUDE conditions, and then any EXCLUDE conditions".
- Set the membership duration and save.
Two builder options change who stays in the list. For dimensions, "At any point" keeps anyone who ever matched (static). Without it, users drop out when they stop matching (dynamic). For metrics and event counts you can use a time window, for example more than 5 events in any 7-day period, instead of a lifetime total.
Limits worth knowing: 100 audiences per standard property (400 on Analytics 360), 500 characters per condition, and after saving you can edit only the name and description. To change the conditions, create a new audience. Archiving is final. Google says it "permanently removes it", and the same name can take up to 48 hours to become free again.
Good starting audiences for a service business are usually simple: visitors who viewed a service or pricing page but did not send an enquiry, people who started a form but did not finish, and past enquirers you want to exclude from prospecting campaigns. All of them rely on the events you already track, so get GA4 set up properly and your own visits filtered out first.
Membership duration
Each audience has a membership duration: how long a user stays in it after they last qualified. Google's builder notes give a maximum of "540 days/18 months from the time user is added to audience last" and say that with Reset on new activity enabled, "the membership audience duration is limited to 14 months". Each time a user meets the conditions again, the clock starts over.
Pick a length that matches the decision you sell. A few weeks suits an abandoned basket. A longer window can suit a service that people research for months. Google also notes that once an audience is exported, "GA retention settings no longer have control. The receiving product will apply their own policies."
Predictive audiences and their thresholds
GA4 can score users with machine learning and let you build audiences on those scores. Google's predictive metrics page lists purchase probability (a user active in the last 28 days making a purchase in the next 7 days), churn probability (a user active in the last 7 days not being active in the next 7), predicted revenue (over the next 28 days) and in-app purchase probability.
The thresholds are high for most small sites. Google says: "In the last 28 days, over a seven-day period, at least 1,000 returning users must have triggered the relevant predictive condition (purchase or churn) and at least 1,000 returning users must not." Model quality also has to hold "over a period of time", and the purchase-based metrics need the purchase or in_app_purchase event, with value and currency sent on purchase. A site that takes enquiries rather than online orders will not get purchase predictions.
If your property qualifies, the predictive audiences page lists five ready-made templates, such as Likely 7-day purchasers (purchase probability above the 90th percentile) and Predicted 28-day top spenders (predicted revenue above the 95th percentile). The builder marks templates you can use as "Ready to use". One warning from Google: if the property later loses eligibility, exported predictive audiences "will not accumulate new users".
Audience triggers
An audience trigger logs a new event when a user joins an audience. You might log a returning_lead event when someone visits three service pages in a week. You can report on that event and mark it as a key event like any other.
Set it up in the audience builder under "Audience trigger". A property can have "up to 20 audience-trigger events". There is an option to fire again "up to one per day" each time membership refreshes. Without it the event fires once, when the user joins. Google also notes "You can't use audience-trigger events as the basis for audience conditions."
Sharing audiences with Google Ads
Audiences reach Google Ads through the product link. Google's connect Google Ads to Google Analytics page sets out what you need:
- Permissions. Editor or administrator on the GA4 property, and administrative access in Google Ads.
- The link itself. In Admin, under Product links, click Google Ads links, then Link. "The option to Enable Personalized Advertising is on by default", and you can turn on auto-tagging in the same step.
- Remarketing switched on in the property. Google says: "For remarketing, at least one of the following must be activated for the property", meaning Google signals or user-provided data collection.
With the link in place and personalized advertising on, Google says your audiences "are automatically shared with Google Ads" and appear in the shared library. You then add them to a campaign or ad group, as described in our remarketing guide. Google also lets people with at least Marketer access create GA4 audiences from inside Google Ads. Those still count towards the property's audience limit.
If you delete the link, Google says "Analytics remarketing audiences stop accumulating new users in unlinked Google Ads accounts." Customer email lists are a separate route through Customer Match, and they do not need GA4.
Why the list is smaller in Google Ads
Expect the Google Ads number to be lower than the GA4 number. Google explains that Analytics counts everyone who meets the conditions, while "Google Ads reports only the subset of those users who are eligible for remarketing". The main reasons Google gives:
- Separate sizes per network. Google Ads keeps different list sizes for Search, YouTube, Gmail and Display & Video 360. For example, a user only counts in a Search list "if they have actively visited Google Search within the preceding 30 days".
- Consent. Visitors who decline ads storage in your cookie banner "are ineligible for remarketing campaigns", even if they still appear in Analytics. The same goes for people who opt out of ads personalization or deny Apple's tracking prompt.
- Your own settings. Events, user properties or regions excluded from ads personalization in GA4 are left out of exports.
- Timing. If you link Google Ads or turn on personalization more than 30 days after creating an audience, Google says the two lists "won't align".
If a list stays empty, Google's troubleshooting page suggests these checks: confirm remarketing is enabled ("Activating Google Signals is a requirement for all Web Remarketing and all App Remarketing"), look at the Google Analytics card under Audience Sources in Google Ads, open the Audiences report in GA4, and loosen an audience that is defined too narrowly.
Also check whether the list is big enough to run. Google Ads sets its own minimum list sizes per network. Our remarketing guide covers those, along with the policies on what you may not target.
A checklist
- Key events tracked and internal traffic filtered before you build anything.
- Marketer role (to create audiences) and Editor role (to link Google Ads) confirmed.
- Audiences created now for campaigns you may run later, since they only fill forward.
- Membership duration chosen to match how long customers take to decide.
- Google Ads linked with personalized advertising on, plus Google signals or user-provided data collection active.
- Consent banner wired to consent mode so that people who agree can be counted.
- Audience report in GA4 and the Audience Sources card in Google Ads checked a few days later.
- Predictive audiences only if the property shows them as Ready to use.
FAQ
Are GA4 audiences retroactive?
No. Google says an audience accumulates users who meet the conditions from the point you create it onward. The copy sent to Google Ads is backfilled with users who matched in the last 30 days, but the audience report in Analytics is not.
How long does a GA4 audience take to populate?
Google says it can take 24 to 48 hours for a new audience to accumulate users, and the same for a new user to be added once they match.
Why is my GA4 audience not showing in Google Ads?
Check that Google Ads is linked with personalized advertising on, that Google signals or user-provided data collection is active in the property, and that the audience is not too narrow. Google Ads also counts only users eligible for remarketing, so the list is smaller there.
Can I edit a GA4 audience after saving it?
Only its name and description. To change the conditions, create a new audience. Archiving an audience removes it permanently.
How many audiences can a GA4 property have?
Up to 100 per standard property and 400 per Analytics 360 property, according to Google.
What do I need for predictive audiences in GA4?
Google requires at least 1,000 returning users who triggered the predicted condition and 1,000 who did not, within a seven-day period in the last 28 days, plus sustained model quality. Purchase predictions also need purchase events with value and currency.
Want tracking and audiences set up before your next campaign? See our website development service and all our services, or email [email protected].