How to Retarget AI-Referred Visitors Using GA4 Audiences and Google Ads

Every week, thousands of visitors arrive at brand websites after asking ChatGPT, Perplexity, or Gemini for a recommendation. These visitors are different from everyone else in your retargeting pool. They didn't stumble in from a banner ad or a broad keyword. They received a specific AI recommendation and chose to click through.

Most of them don't convert on the first visit. Most high-intent buyers don't — they research, compare, and return. But almost no brand is running retargeting campaigns specifically designed for this audience.

This guide explains exactly how to build that system: five GA4 audience segments, the Google Ads campaign structure to match, bid strategy, ad copy approach, and how to diagnose the funnel issues that kill retargeting ROI before you spend a dollar.

https://cdn.sanity.io/images/en9d2pb2/production/de333cabeb9c6e85dc37a7eb7723849d4b5c7748-1200x630.png

If you haven't set up AI traffic tracking yet, start with how to track AI referral traffic in GA4 first. This guide picks up where that one ends.

---

Why AI-Referred Visitors Are Your Best Retargeting Audience

Three things make AI-referred visitors fundamentally different from any other retargeting segment you're running.

They received an AI recommendation. A user didn't passively scroll past your ad. They asked an AI system a specific question — "what's the best project management tool for agencies" or "which protein powder has the best amino acid profile" — and the AI named your brand. That's a trust transfer unlike anything a display ad can manufacture.

They arrived with validated intent. By the time an AI-referred visitor lands on your site, they've already done the research phase. The AI did it for them. They're further down the consideration funnel than almost any other traffic source.

They came with trust. People trust AI recommendations more than they trust ads. Multiple surveys in 2025 found that 60-70% of users consider AI-suggested products more credible than sponsored search results. When that visitor bounces without converting, you're not retargeting a skeptic — you're retargeting someone who was already warm.

The gap between intent and conversion is where retargeting earns its keep. For AI-referred visitors, that gap is smaller and the path to closing it is cleaner.

---

Why Most Brands Aren't Retargeting This Audience Yet

Two structural problems explain the attribution gap.

The audience isn't defined. Standard remarketing setups target "all site visitors" or "product page visitors." These are coarse segments that mix AI-referred visitors in with people who bounced in 4 seconds from a Reddit thread. Nobody has pulled AI-referred visitors out as their own audience — so nobody is targeting them with messaging calibrated for their actual context.

The attribution data isn't being used. Most marketing teams check their GA4 AI traffic numbers once a quarter if at all. The data exists. The sessions are being tracked. But the operational connection between "this visitor came from ChatGPT" and "let's retarget them differently" hasn't been built.

Building that connection is exactly what this guide does.

---

Prerequisites Before You Start

Two things need to be in place before any of this works.

GA4 linked to Google Ads. If your GA4 property and Google Ads account aren't linked, audiences you build in GA4 won't appear in Google Ads for targeting. To link: GA4 Admin → Google Ads Links → Create link → select your Google Ads account → confirm. Audience sync begins within 24-48 hours.

Minimum 100 AI sessions per month. Google Ads requires a minimum of 100 active users in a remarketing list to activate Search and Gmail campaigns, and 1,000 users for Display and YouTube. If your site currently gets fewer than 100 AI-referred sessions monthly, build the broad audience now and let it accumulate — don't wait. Use the AI traffic decoder to monitor current AI session volume across all platforms.

---

Step 1: Create GA4 AI Audience Segments

You need five audiences. Each one maps to a different stage of the buyer journey and enables different campaign strategies.

To create each audience in GA4: Admin → Audiences → New Audience → Create a custom audience.

Audience Configuration Table

Audience Name	Conditions	Lookback Window	Primary Use
AI_Referred_All	Session source contains chatgpt.com OR perplexity.ai OR gemini.google.com OR claude.ai OR copilot.microsoft.com	30 days	Broad retargeting, brand awareness
AI_High_Intent	AI source (above) AND engaged session (session duration >30s OR pageviews ≥2)	30 days	Mid-funnel display and YouTube
AI_Product_Viewer	AI source AND page_location contains /product OR /pricing	14 days	Dynamic product ads, RLSA
AI_Non_Converter	AI source AND session occurred AND purchase event NOT triggered	30 days	Conversion-focused Search and Display
AI_Cart_Abandoner	AI source AND add_to_cart event fired AND purchase event NOT triggered	7 days	Highest-urgency retargeting

Audience 1: AI_Referred_All

This is your base layer — every visitor from any major AI platform in the last 30 days.

Configuration:

• Condition: Session source contains chatgpt.com
• OR: Session source contains perplexity.ai
• OR: Session source contains gemini.google.com
• OR: Session source contains claude.ai
• OR: Session source contains copilot.microsoft.com
• Lookback window: 30 days
• Membership duration: 30 days

This audience feeds brand awareness campaigns and provides the population from which all other segments are carved.

Audience 2: AI_High_Intent

Visitors who came from AI and then demonstrated genuine engagement — not just a 3-second bounce.

Configuration:

• Condition: All conditions from AI_Referred_All (AI source)
• AND: Session duration greater than 30 (seconds)
• OR within AND group: Session engaged equals true (catches GA4's engaged session flag which requires 10s+ AND interaction OR 2+ pageviews)
• Lookback window: 30 days

This segment filters out the bounce traffic and isolates the visitors who actually read your content, scrolled through your product pages, and spent meaningful time evaluating your offer.

Audience 3: AI_Product_Viewer

Visitors who came from AI and landed on or navigated to product or pricing pages — the highest purchase-intent signal available before cart.

Configuration:

• Condition: All AI source conditions
• AND: Page location contains /product OR Page location contains /pricing OR Page location contains /shop
• Lookback window: 14 days (shorter window = fresher, higher-intent audience)
Adjust the page path conditions to match your actual URL structure. For SaaS, this is typically /pricing. For ecommerce, it's /products or /collections. For B2B, it's /solutions or /features.

Audience 4: AI_Non_Converter

Everyone who came from AI and left without converting. The largest and most actionable segment for conversion-focused campaigns.

Configuration:

• Condition: All AI source conditions
• AND: Event session_start (to confirm a session occurred)
• AND NOT: Event purchase (or your primary conversion event)
• Lookback window: 30 days
For B2B/SaaS: replace purchase with form_submit, demo_booked, or calendly.event_scheduled.

Audience 5: AI_Cart_Abandoner

The highest-urgency segment. These visitors received an AI recommendation, clicked through, added a product to their cart, and left. Every hour they remain unconverted is revenue at risk.

Configuration:

• Condition: All AI source conditions
• AND: Event add_to_cart was triggered
• AND NOT: Event purchase was triggered
• Lookback window: 7 days

These users need to be reached within 24-48 hours with friction-reducing messaging (free shipping, easy returns, social proof). A 7-day window keeps the list fresh and high-intent.

https://cdn.sanity.io/images/en9d2pb2/production/bf46a3b0a15610f5f2e001fa0a4e3c009dcae17e-1200x630.png

---

Step 2: Link GA4 to Google Ads

If you skipped this in the prerequisites, here's the exact process.

• In GA4, navigate to Admin (bottom-left gear icon)
• Under Property, select Google Ads Links
• Click Create link
• Choose your Google Ads account from the list (you must have admin access to both)
• Enable Personalized advertising (required for remarketing audiences)
• Enable Auto-tagging if it isn't already active
• Review and submit

Audiences created before the link will not backfill. Audiences you create after the link will begin syncing within 24-48 hours and appear in Google Ads under Tools → Audience Manager → Your data segments.

For GA4 custom reports for AI traffic that complement your audience setup, see the linked guide.

---

Step 3: Wait for Audience Population

Patience is required here. GA4 audiences don't appear in Google Ads until they meet minimum size thresholds.

Campaign Type	Minimum List Size	Typical Population Timeline
Search (RLSA)	1,000 users	2-4 weeks for most sites
Gmail	100 users	7-14 days
Display	100 users	7-14 days
YouTube	1,000 users	2-4 weeks
Performance Max	1,000 users	2-4 weeks

Note: Google Ads documentation states 100 users for Display and Gmail, but in practice many accounts see lists become "active" only after 100-500 users depending on account history and privacy thresholds.

During the population window:

• Do not delete and recreate audiences — this resets the clock
• Monitor daily in GA4 under Admin → Audiences to watch user counts grow
• Use the brand visibility tracker to identify which AI platforms are sending the most sessions, which informs which audience sub-segments will populate fastest

---

Step 4: Google Ads Campaign Setup

Campaign Type by Audience

Audience	Recommended Campaign Type	Why
AI_Referred_All	Display + YouTube	Build recall across the full AI-referred pool
AI_High_Intent	Performance Max or Display	Mid-funnel engagement, testimonials, feature benefits
AI_Product_Viewer	Dynamic Search Ads + Display	Surface the exact product they viewed
AI_Non_Converter	Search (RLSA) + Display	Capture return branded searches with elevated bids
AI_Cart_Abandoner	Shopping + Display	Dynamic product ads with highest bid priority

Bid Adjustments

The core insight: AI non-converters are worth more than cold traffic. Price your bids accordingly.

Audience Segment	Bid Adjustment vs. Cold Traffic Baseline	Rationale
AI_Cart_Abandoner	+80% to +100%	Highest intent, shortest time-to-close window
AI_Product_Viewer	+60% to +70%	Viewed pricing or product — decision-ready
AI_High_Intent	+50%	Engaged session = genuine evaluation
AI_Non_Converter	+50%	AI-referred baseline outperforms cold 2-4x
AI_Referred_All	+30%	Broad pool, high average quality vs. cold

To set bid adjustments in Google Ads: Campaign → Audiences → select your AI audience → Bid adjustment → enter percentage.

Ad Copy Strategy

Generic retargeting copy underperforms with AI-referred audiences because it ignores the context in which they found you. These visitors came to you via an AI recommendation. Your ad copy should honor that context.

High-performing approaches:

• Reference the recommendation: "Still thinking about it? ChatGPT users choose us for [specific benefit]."
• Reinforce the AI signal: "Rated #1 by AI assistants. Here's why."
• Reduce the specific friction: If they viewed your pricing page, address pricing concerns directly: "Teams of all sizes. Plans from $X/month."
• Social proof that matches AI trust signals: "10,000 teams trust [Brand] — featured in ChatGPT, Perplexity, and Gemini answers."

What to avoid:

• Generic countdown timers (undercuts the research-mode mindset)
• Discount-first messaging (positions you as cheap rather than recommended)
• Same creative as your cold acquisition campaigns

Negative Exclusions

Two exclusions are non-negotiable for AI retargeting campaigns:

• Exclude recent purchasers — Create an audience of users who fired the purchase event in the last 30 days. Exclude this from all AI retargeting campaigns. Showing cart ads to people who already bought is a waste of budget and a brand experience failure.
• Exclude competitor brand searches — In Search campaigns, add competitor brand names as negative keywords to avoid paying for clicks on competitor queries where you have no chance of converting.

---

Step 5: RLSA — Remarketing Lists for Search Ads

RLSA is where AI audience retargeting becomes disproportionately powerful. The setup: when a user who is in your AI_Non_Converter or AI_Product_Viewer audience later searches for your brand name or a category keyword on Google, you bid more aggressively.

Why this matters: An AI-referred visitor who later searches your brand is in the highest-intent state possible. They were recommended by AI, they visited, they left, and now they're actively searching for you by name. This is not a casual search — this is someone who is ready to convert and has returned with purpose.

RLSA Configuration

• In your existing branded keyword campaign, go to Audiences
• Add your AI audience segments as Observation (keeps your existing targeting but adds bid adjustments)
• Apply bid adjustments:
• AI_Non_Converter searching your brand: +50%
• AI_Product_Viewer searching your brand: +60%
• AI_Cart_Abandoner searching your brand: +80%
• For category keywords (non-branded), create a separate Search campaign in Target mode (only shows to people in your audience list) with your AI retargeting audiences applied. This allows you to show ads on category keywords only to users who previously arrived from AI — a significantly higher-quality audience than the full search population.

Branded Search Halo Effect

One documented phenomenon: AI recommendations increase branded search volume. Users who see your brand recommended by Perplexity often search your brand name on Google before clicking. This is the "branded search halo" effect described in the how to track AI referral traffic in GA4 guide.

RLSA lets you intercept these searches with elevated bids and specific messaging that acknowledges the AI touchpoint in the buyer journey — closing a loop that most attribution models leave invisible.

---

Performance Benchmarks: What to Expect

Conversion Rate

Audience Segment	Expected CVR vs. Cold Retargeting	vs. Organic Search
AI_Cart_Abandoner	4-6x higher	2-3x higher
AI_Product_Viewer	3-4x higher	1.5-2x higher
AI_High_Intent	2-3x higher	Comparable to 1x
AI_Non_Converter	2-4x higher	1-1.5x higher
AI_Referred_All	1.5-2x higher	Slightly below

CPA

Metric	AI Retargeting vs. Cold Retargeting
CPA reduction	30-50% lower
ROAS improvement	2-3x higher
Payback period	Typically 7-14 days vs. 30-45 days cold

Timeline to Results

Phase	Timeline	Milestone
Audiences created	Day 0	Tracking starts
GA4 → Google Ads sync	Day 1-2	Audiences visible in Ads
Search/Gmail eligible	Day 7-14	100+ users in list
Display/YouTube eligible	Day 14-30	1,000+ users in list
Statistically meaningful data	Day 30-45	Enough conversions to optimize

For sites with strong AI visibility (2,000+ AI sessions/month), lists qualify faster and performance data appears sooner. Use the AI traffic decoder to monitor weekly AI session volume so you know exactly when each audience threshold will be hit.

---

The B2B Version: SaaS and Lead Generation

Everything above applies to B2B and SaaS — but the conversion events are different.

Replace purchase with:
• form_submit (contact forms, demo request forms)
• demo_booked (custom event after Calendly or Chili Piper booking)
• calendly.event_scheduled (if you're tracking Calendly via GTM — see the fix below)
• video_complete_50 (for product demo videos — 50% watched is a strong intent signal)

B2B Audience Definitions

Audience	Conditions	Priority
AI_Visited_Pricing	AI source AND page_location contains /pricing	High
AI_Visited_Demo	AI source AND page_location contains /book-demo OR /demo	Highest
AI_Watched_Demo_Video	AI source AND video_complete event with percent ≥50	High
AI_Non_MQL	AI source AND session occurred AND form_submit NOT triggered	Medium

For B2B RLSA, bid +50-70% on branded keywords for users who visited your /pricing or /book-demo pages via AI. These are the warmest possible branded search queries you will ever see.

---

Diagnosing Broken Funnels Before You Retarget

Before you launch any retargeting campaign, run this audit. Retargeting traffic to a broken landing page or untracked conversion flow amplifies the problem — you spend more to get the same zero conversions.

The Most Common Funnel Breaks

1. Form tracking is broken. If your GA4 shows high sessions on /contact or /book-demo but zero form_submit or form_start events, your form events aren't firing. This means your AI_Non_Converter audience is including people who actually submitted forms — a poisoned audience. 2. Calendly iframe not tracked. This one is endemic. In one audit, a B2B SaaS site's /book-demo page received 3,250 visits over 90 days but recorded 0 form_start events. The form was a Calendly embed inside an iframe — an entirely separate domain from the parent page. GA4 has no visibility into iframe events by default. The fix: Use Google Tag Manager to listen for the calendly.event_scheduled postMessage that Calendly fires to the parent window when a booking is completed.

GTM Custom HTML tag:

Trigger: All Pages. Then create a GA4 Event tag in GTM that fires on the calendly_booking_completed dataLayer event and sends a demo_booked event to GA4. 3. Mobile sessions under 5 seconds. If your /pricing or /book-demo page shows average session duration under 5 seconds on mobile, the page likely isn't loading correctly on mobile devices — either the form widget isn't rendering or the CTA button isn't visible without scrolling. Retargeting mobile users to this page will produce the same failure at higher cost.

Pre-Retargeting Checklist

Check	How to Verify	Action if Broken
Form events firing	GA4 Realtime → Events → submit a test form	Fix GTM trigger or GA4 event tag
Calendly tracked	GTM Preview → trigger calendly_booking_completed	Add postMessage listener in GTM
Mobile page load	GA4 → Tech → Mobile Overview → avg session duration	Fix mobile layout, test on real device
Conversion event in GA4	Admin → Conversions → verify event is marked	Mark correct event as conversion
GA4 → Ads link active	GA4 Admin → Google Ads Links → status = active	Relink if disconnected

Only run retargeting campaigns after each of these checks passes. If your conversion tracking is broken, your audience definitions are wrong, your bid optimization signals are corrupted, and your reported ROAS is meaningless.

For a complete audit of your AI traffic measurement setup, the AI traffic decoder surfaces attribution gaps and broken funnel steps automatically. The prompt intelligence feature shows you exactly which AI queries are sending visitors to your site — critical context for writing retargeting ad copy that resonates.

---

Putting It All Together: The AI Retargeting Stack

Layer	Tool	Purpose
Traffic identification	GA4 + AI traffic decoder	Tag all AI-referred sessions
Audience building	GA4 Audiences	Segment by behavior and funnel stage
Ad serving	Google Ads	Deliver targeted campaigns
Bid intelligence	RLSA + smart bidding	Maximize value from high-intent searches
Funnel monitoring	GA4 + GTM	Ensure conversion events fire correctly
AI visibility	Brand visibility tracker	Monitor which AI platforms cite your brand
Prompt coverage	Prompt intelligence	Know what queries drive your AI traffic

The brands that win in AI-influenced commerce aren't just getting recommended — they're closing the loop from recommendation to retargeting to conversion. That full-funnel view is what separates AI traffic from a vanity metric to a real acquisition channel.

---

Next Steps

If you're new to AI traffic measurement, start with the how to track AI referral traffic in GA4 guide to establish your baseline attribution. Then return to this guide to build out the retargeting layer.

For a complete picture of your brand's AI visibility — which platforms mention you, in which contexts, and which competitors are being recommended instead of you — the brand visibility tracker provides real-time monitoring across ChatGPT, Perplexity, Gemini, and Claude.

To understand the strategic framework that gets your brand recommended by AI in the first place, read our guide to answer engine optimization.

Book a demo to see how Asva AI handles the full stack — from tracking AI-referred sessions to building audiences to monitoring the AI visibility that drives them.