Google Analytics 4 (GA4) Audiences: The 'Predictive' Targeting Weapon (2024)

Old Google Analytics (UA) was about History: "What happened yesterday?" Google Analytics 4 (GA4) is about Prediction: "What will happen tomorrow?"

The killer feature of GA4 for advertisers is Predictive Audiences. Google's machine learning looks at user behavior (scroll depth, time on site, specific clicks) and assigns a "Probability Score" to every user.

You can then sync these audiences to Google Ads to run sniper-focused campaigns. This is how you lower CPA by 30%.

Part 1: The "Predictive" Segments

GA4 offers three native predictive metrics out of the box (if you have enough data).

1. Likely to Purchase (next 7 days):

   • These users are hot. They viewed the pricing page, added to cart, read the shipping policy.
   • Strategy: High-Bid Retargeting. Show them a distinct "10% Off to finish your order" ad.

2. Likely to Churn (next 7 days):

   • These are active users who are showing signs of leaving (visiting the "Cancel Subscription" page, reduced login frequency).
   • Strategy: Retention Campaign. Show them a "We miss you" or "New Feature Update" ad.

3. Predicted Revenue (High Spenders):

   • Users likely to spend top-tier amounts.
   • Strategy: Lookalike Seed. Use this audience to build a standard Google Ads "Similar Segment" to find new whales.

Part 2: Connectivity (Link GA4 to Ads)

You must link the properties.

1. GA4: Admin → Product Links → Google Ads Links.
2. Google Ads: Tools → Manager → Linked Accounts → Google Analytics (GA4) & Firebase.
3. Activate: Ensure "Personalized Advertising" is turned ON in the link settings.

Once linked, your GA4 audiences appear in the Google Ads Audience Manager automatically.

Part 3: Creating Custom Dimensions

Predictive is great, but Custom Dimensions vary by business.

SaaS Example:

• You track an event called viewed_api_docs.
• Users who view API docs are likely Developers (Influencers), not CEOs (Buyers).
• Action: Create an audience in GA4: "Include users who fired viewed_api_docs".
• Targeting: Exclude this audience from your "Request Demo" campaigns (so you don't pay for devs). Target them with "Start Free Trial" campaigns.

Ecommerce Example:

• You track purchase_value.
• Action: Create an audience: "Whales" (LTV → $500).
• Targeting: Bid only on generic keywords (e.g., "luxury shoes") if the user is in the "Whales" audience (via Observation/RLSA).

Part 4: The 7-Day vs 30-Day Debate

Membership duration matters.

• 7-Day Windows: High urgency. If they don't buy in 7 days, they are gone. High CPAs are acceptable here.
• 30-Day Windows: Nurturing. Good for B2B.
• 540-Day Windows: Brand awareness. Keep top-of-mind.

Summary

Don't just use the default "All Visitors" list. It is lazy and inefficient.

1. Verify GA4 is linked.
2. Check the "Predictive" tab in the Audience Builder.
3. If eligible, launch a "Likely to Purchase" campaign today.

Let the AI identify the buyers within the browsers.