Short answer
A GA4 audience is built under Audiences in Admin by defining conditions or a sequence. To use it in Google Ads, the property must be linked to Google Ads, personalised advertising must be on, Google signals or user-provided data collection must be active, and the visitor must have given ad consent.
Say a shoe shop builds an “abandoned carts” audience in GA4, adds it to a campaign in Google Ads and checks the list the next day: empty. The audience definition was fine; one of the settings between GA4 and Google Ads was off. The question of how to build GA4 audiences and use them in Google Ads has two halves: defining the audience correctly, and keeping on the settings that let it reach the ad account.
We covered the ad side (exclusions, frequency, the message of the second ad) in our remarketing article and the GA4 property itself in the setup checklist. This article covers the link between the two: building the audience in GA4 and sending it on its way.
What an audience is in GA4
Google’s audience help page describes an audience as a way to segment users by dimensions, metrics and events into groups that matter to your business. As new data arrives, memberships are re-evaluated; a user who no longer meets the criteria is removed from the audience.
Two points are worth knowing up front. An audience collects users from the moment it is created, and a new audience can take 24 to 48 hours to fill. Creating one also needs at least the Marketer role on the property. The management table holds two predefined audiences: All users and Purchasers.
Step 1: Start the audience
In Admin, go to Audiences under Data display, then New audience. Google offers three routes:
- Custom audience: you define every parameter yourself.
- Template: you modify a partly configured audience; the demographics template, for example, holds age, gender, language and location dimensions.
- Suggested audience: a ready-made audience based on your industry category and the events recommended for it, used as it is or modified.
According to Google’s suggested audiences page, you need to collect the events and parameters an audience is based on for it to fill. Without an add-to-cart event, an “abandoned carts” audience will not fill. That is why we suggest fixing event names first in the measurement plan.
Step 2: Build the conditions
Every condition has a scope: across all sessions, within the same session, or within the same event. An audience can hold up to 10 conditions. A condition rests on one of three kinds of data:
- Dimensions. Age or page address, for example. If you tick “At any point”, a user stays in the audience even if they met the condition only once (static evaluation). If you leave it unticked, users join when they meet the condition and leave when they no longer do (dynamic evaluation).
- Metrics. Lifetime value, for example. Time-windowed conditions such as “in any 7-day period” are possible too.
- Events. Any event you collect; when the scope is across all sessions, you can also add a time window in days to the event count (event_count) parameter.
You keep people out with an exclude group. According to Google, users are evaluated against the include conditions first and the exclude conditions second; a user who meets an exclude condition is not added. You can exclude temporarily or permanently. Removing purchasers from a cart audience is the typical use.
Step 3: Add a sequence if you need one
A sequence sets the order in which conditions must be met. You define whether a step is “indirectly followed by” the next one (any time afterwards) or “directly followed by” it (immediately afterwards), and you can set a time limit for the whole sequence. Example: first a product page view, then add to cart, with purchase excluded.
A sequence is stricter than a plain “and”. A user who did both things but in the wrong order does not enter. On a small site this can make the audience narrower than you want, so we suggest starting with plain conditions.
Step 4: Choose the membership duration
Membership duration decides how long a user stays in the audience. According to Google it is at most 540 days, counted from the last time the user was added to the audience. Each time the user meets the criteria again, the duration resets to its full value. For audiences exported to other products, GA4’s retention settings no longer apply; the receiving product applies its own policies.
We suggest choosing the duration to fit the product’s buying cycle. For something bought on impulse, showing a cart reminder months later makes no sense, so keep the duration short. For an expensive service that people think about for a long time, a longer duration can be reasonable.
The limits
We have gathered the limits from Google’s pages in one table.
| Limit | Value |
|---|---|
| Audiences per standard property | 100 |
| Audiences per 360 property | 400 |
| Conditions in one audience | up to 10 |
| Length of one condition | 500 characters |
| Membership duration | up to 540 days |
| A new audience filling | 24 to 48 hours |
| Pre-population of the Google Ads list | up to 30 days of data |
| Editable after creation | name and description only |
| Reusing the name of an archived audience | you may have to wait up to 48 hours |
| Google Ads links per property | up to 400 |
Archiving is permanent: according to Google, you cannot access or restore an archived audience. If a condition is wrong you cannot edit it; you duplicate the audience and build a new one. When you reach the limit, archive the old audiences.
Step 5: Predictive audiences and their eligibility requirements
A predictive audience is one with at least one condition based on a predictive metric. Google’s predictive audiences page gives users likely to purchase in the next 7 days as an example. The suggested “Likely 7-day purchasers” audience includes users whose purchase probability is above the 90th percentile.
Prediction is not available on every property. Google’s predictive metrics page lists these requirements:
- Within the last 28 days, over a seven-day period, at least 1,000 returning users must have triggered the relevant condition (purchase or churn) and at least 1,000 returning users must not have.
- Model quality must be sustained over a period of time.
- For purchase probability and predicted revenue, the property must send the purchase and/or in_app_purchase event, and the value and currency parameters must be collected with purchase.
Predictions are generated once a day per active user for each eligible model. If quality falls below the threshold, Google stops updating them, and exported predictive audiences stop accumulating new users. You can see eligibility while creating an audience, under the Predictive tab of the suggested audiences, from the “Ready to use” label. On a small site these thresholds may not be met; in that case, carry on with audiences built from plain conditions.
Step 6: Link to Google Ads and check what must be on
An audience sitting in GA4 is invisible to the ad account. According to Google’s linking page, you create the link under Product links, Google Ads links, in Admin. You need at least the Editor role on the GA4 property and admin access in Google Ads. “Enable Personalized Advertising” is on by default; it must stay on for audiences to flow to Google Ads.
For remarketing, Google signals or user-provided data collection must also be active. The page states this explicitly. If you delete the link, audiences stop accumulating new users in Google Ads.
Once linked, the audiences appear in the shared library in Google Ads and can be added to a campaign or an ad group. According to Google, the list in Google Ads is pre-populated with up to 30 days of data, where the data is available.
Consent and audience size: why the two numbers differ
An audience size in GA4 is the number of users who meet the criteria. The number in Google Ads shows only the part that is eligible for advertising. Google’s audience size page lists the main reasons:
- A visitor who declined ad data storage but accepted analytics storage can still appear in GA4 reports, but is not counted in linked advertising products.
- Personalised advertising can be switched off at the level of the property, event, region, link and the user’s Google Account. If it is off, the user does not enter the Google Ads list.
- A Search remarketing list only includes people who actively used Google Search in the preceding 30 days.
- GA4 reports are not backfilled, but the Google Ads list is. In the first days the numbers therefore diverge.
- The summary card in the audience builder is an estimate based on the last 30 days.
So two numbers that disagree are often not an error. Set up the consent banner and consent mode in the order described in our cookie consent article. Keeping visitors who decline out of the list is a legal matter as much as a technical one, and you should assess it with your legal adviser.
A starting set of audiences for a small site
Opening an audience for every page wastes the limit of 100 audiences. The table below is the simple starting set we suggest; the event names in the cart and checkout rows come from the “Cart abandoners” and “Checkout starters” audiences on Google’s suggested audiences page.
| Audience | How to build it in GA4 | What it is for |
|---|---|---|
| Cart abandoners | add_to_cart event present, purchase event absent (exclude) | The group with the clearest intent to buy |
| Checkout starters | begin_checkout present, purchase absent | A message that removes the obstacle at the payment step |
| Quote requesters who stopped | Price or service page viewed, generate_lead absent | Stands in for the cart on service sites |
| Purchasers | The predefined “Purchasers” audience | Excluding from ads and for repeat purchases |
| All users | The predefined audience | A fallback until the others fill |
Three or four audiences are enough. Add a predictive audience only if you see “Ready to use”. The events the audiences rest on must be verified first; we do this work, starting with the measurement plan, in our data and CRO service. Without a purchase event, the exclusion does not work.
Common mistakes
- Building the audience without checking the link. With no link, or with personalised advertising off, the list never fills.
- Leaving Google signals off. For remarketing, Google signals or user-provided data collection must be active.
- Not excluding purchasers. Someone who has ordered keeps seeing the cart reminder.
- Making the condition too narrow. Sequences and extra conditions shrink the audience; we explain Google Ads’ list-size conditions in the remarketing article.
- Leaving the membership duration unconsidered. A duration that does not fit the buying cycle chases old interest with ads.
- Using up the audience limit early. Archiving is permanent; do not open audiences you will not use.
- Taking a mismatch between the two numbers for an error. The difference mostly comes from consent and eligibility rules.
Common questions
When does a GA4 audience appear in Google Ads?
According to Google, a new audience can take 24 to 48 hours to start collecting users. The list in Google Ads is also pre-populated with users from up to the last 30 days. If the link or permission settings are incomplete, the list never fills.
Can I change the conditions after creating the audience?
No. According to Google, only the name and description can be edited after creation. To change a condition, duplicate the audience and build a new one, then archive the old one if you need to.
Can I use predictive audiences on a small site?
Only if the property meets the requirements. Google asks for at least 1,000 returning users who trigger the condition and 1,000 who do not, over a seven-day period. Low-traffic sites often fall short; check for the “Ready to use” label.
One thing you can do today
Open the Google Ads links in Admin in GA4 and confirm that personalised advertising is on. Then check that Google signals or user-provided data collection is active on the property. If all of that is in place, build a single audience: users with add_to_cart who do not have purchase. Three days later, write down the GA4 and Google Ads numbers side by side; if they differ, rule out the reasons above one by one.
Sources
- Google Analytics Help: Create, edit, and archive audiences
- Google Analytics Help: Predictive audiences
- Google Analytics Help: Predictive metrics
- Google Analytics Help: Connect Google Ads to Google Analytics
- Google Analytics Help: Audience size differences between Google Analytics and Google Ads
- Google Analytics Help: Suggested audiences
This article is general information, not legal, financial or technical advice. Platforms, their rules and the law change over time, so some details may have changed since 15 November 2026 or contain small errors. Before you make a decision for your business, check the current official sources or write to us.