Why Google Analytics Can't Track AI Commerce (And What to Do About It)

For more than a decade, Google Analytics has been the default measurement tool for e-commerce businesses. The migration to GA4 brought new capabilities, enhanced machine learning features, and a more flexible event-based data model. Yet as AI commerce emerges as a dominant force in product discovery, GA4 finds itself fundamentally unsuited to answer the questions that matter most.

This isn't a criticism of Google Analytics—it's a recognition that GA4 was designed for a different era. The tool was architected to measure web traffic, user behavior on owned properties, and conversion paths that flow through traditional digital touchpoints. AI commerce operates on entirely different principles.

Understanding why Google Analytics can't track AI commerce—and accepting this limitation—is the first step toward building a measurement approach that actually captures AI's influence on your business.

The GA4 Design Assumptions

Google Analytics 4 rests on several foundational assumptions that made perfect sense when the platform was designed but create fundamental blind spots in the AI commerce context.

Assumption 1: Users Visit Your Properties

The most basic assumption underlying GA4 is that users you want to measure visit properties you control—your website, your app, your digital destinations where you've installed tracking code.

AI commerce inverts this model entirely. When a consumer asks ChatGPT "what's the best running shoe for flat feet?" and receives a recommendation for a specific product, the critical moment of influence happens on OpenAI's property, not yours. When Google's AI Overview synthesizes product information from across the web and presents a curated answer, the user may form strong purchase intent before ever touching your website.

GA4 can measure what happens when users finally arrive at your site. It cannot see the AI interactions that shaped their intent before arrival. This distinction matters enormously because AI influence often occurs at the highest-leverage moments of the purchase journey—the moments when consumers are forming preferences and narrowing consideration sets.

Assumption 2: Influence Creates Trackable Events

GA4's event-based model is powerful but depends on user actions generating measurable events—page views, clicks, scroll depth, video plays, form submissions, transactions. The system excels at capturing granular user behavior and synthesizing it into coherent journeys.

AI commerce influence frequently creates no trackable events whatsoever. A consumer who reads a ChatGPT recommendation, mentally notes the product, and purchases it three days later through a different channel has experienced AI influence that GA4 cannot detect. The conversation with the AI assistant exists entirely outside your measurement infrastructure.

Even when AI surfaces do drive direct traffic, the referral data is often obscured. Many AI integrations strip or obfuscate referral information. Users who copy product names from AI recommendations and search for them directly appear as organic or direct traffic—indistinguishable from users who discovered products through other means.

Assumption 3: Sessions Define User Journeys

GA4's journey analysis centers on sessions—defined periods of user activity on your properties. The platform excels at understanding what users do within and across sessions, identifying drop-off points, and optimizing owned experiences.

AI commerce shatters the session paradigm. A user might have a conversation with an AI assistant, receive product recommendations, and not visit your website for days or weeks—if they visit at all. When they do purchase, the session in which they convert reveals nothing about the AI influence that actually drove the decision.

Cross-device behavior compounds this problem. A user might ask ChatGPT for recommendations on their phone during a commute, later browse on their laptop, and ultimately purchase on a tablet. GA4 struggles to stitch these fragmented journeys together even when all touchpoints occur on your properties. When the critical first touchpoint happens on an AI platform you don't control, journey reconstruction becomes essentially impossible.

What GA4 Can Track (And What It Can't)

Understanding GA4's capabilities and limitations in the AI commerce context requires distinguishing between what the platform can theoretically measure and what it actually reveals about AI influence.

What GA4 Can Track

Post-click behavior: When users do arrive at your site from AI surfaces (with preserved referral data), GA4 can track their subsequent behavior effectively—pages viewed, products examined, cart additions, purchases.

Aggregate traffic patterns: GA4 can identify unusual patterns in traffic or conversion that might correlate with AI commerce activity—though it cannot establish causation.

Direct AI referrals (sometimes): In cases where AI platforms preserve referral data and drive direct traffic, GA4 can attribute sessions to AI sources. However, this represents a small fraction of AI commerce influence.

Conversion outcomes: GA4 accurately tracks conversions that occur on your properties, regardless of the influence that drove them—you just can't attribute those conversions to AI.

What GA4 Cannot Track

Conversational AI interactions: Conversations between users and AI assistants about your products are completely invisible to GA4. You cannot see what questions users ask, what recommendations they receive, or how AI systems describe your products.

AI-influenced intent formation: The critical moments when consumers form purchase intent through AI interactions generate no measurable signals in GA4. By the time users reach your site, intent has already been shaped by forces you cannot observe.

Competitive AI visibility: GA4 tells you nothing about how AI systems view your products relative to competitors. You cannot assess whether AI assistants recommend your products, competitors' products, or neither.

Product data interpretation: GA4 cannot reveal how AI systems interpret your product information—whether titles are parsed correctly, descriptions are understood, or key attributes are extracted and utilized in recommendations.

Attribution to AI influence: Even when AI demonstrably drives purchases, GA4 typically cannot attribute that revenue to AI surfaces. The influence happens before the measurable session begins.

The Referral Attribution Gap

One of the most frustrating limitations for e-commerce teams is GA4's inability to properly attribute AI referrals. This gap has several dimensions.

Obscured Referral Data

Many AI platform integrations intentionally or unintentionally obscure referral information. When users click links within ChatGPT responses, the referral data that would identify ChatGPT as the traffic source is often stripped or replaced with generic referrers.

This means AI-driven traffic frequently appears in GA4 as:

• Direct traffic: Users appear to have typed your URL directly
• Organic search: If users search for products they learned about from AI
• Referral from unexpected sources: Intermediate redirects may replace AI platform referrers

Without reliable referral attribution, GA4 cannot distinguish AI-influenced visitors from the general traffic population.

The Click Dependency Problem

GA4's attribution model requires clicks to create trackable paths. But much AI commerce influence never generates clicks to your properties at all:

• Users may receive recommendations and mentally note them for future purchase
• AI assistants may describe products in ways that influence preference without linking
• Voice-based AI interactions typically don't generate any web traffic
• Users may convert through offline channels after AI research

This click dependency means GA4 systematically underestimates AI influence by ignoring non-click influence entirely.

Multi-Touch Complexity

Even when AI surfaces do drive trackable visits, GA4's multi-touch attribution struggles to weight AI influence appropriately. AI often operates at the top of the funnel—creating awareness and shaping consideration—but GA4's attribution models may credit later touchpoints that captured demand AI created.

A user who discovers your product through ChatGPT, later sees a retargeting ad, and converts through an email link will likely show email or paid social as the conversion driver in GA4. The AI influence that initiated the entire journey receives no attribution credit.

The Session-Based Limitation

GA4's session-based measurement creates additional blind spots specific to AI commerce patterns.

Extended Consideration Cycles

AI commerce often initiates long consideration cycles. Users researching significant purchases may consult AI assistants multiple times over days or weeks, gradually refining preferences and building toward purchase.

GA4 treats these extended journeys as disconnected sessions with no understanding of the AI interactions occurring between website visits. The platform sees isolated data points when the actual user journey is continuous but happens primarily on AI platforms.

Cross-Platform Research

Modern consumers move fluidly across platforms and devices. AI research might happen on a phone, followed by desktop browsing, followed by tablet purchase. GA4's user stitching capabilities help somewhat, but cross-platform identity resolution remains imperfect—and completely fails for AI interactions that occur off your properties.

Zero-Session Conversions

Perhaps most critically, significant AI commerce influence results in what might be called "zero-session conversions"—purchases where AI influence was decisive but the user never visited your website before purchasing.

This happens when:

• Users purchase through marketplace listings after AI recommendation
• AI assistants provide purchase links to retailer sites
• Users call or visit physical stores based on AI research
• Voice commerce converts directly within AI platforms

These conversions represent real revenue influenced by AI that generates zero sessions in GA4.

Workarounds and Their Limitations

Enterprising e-commerce teams have attempted various workarounds to capture AI commerce signals within GA4. While these approaches have some value, their limitations should be clearly understood.

UTM Parameter Campaigns

Some teams create custom landing pages with UTM parameters specifically for AI surfaces, hoping to capture AI referrals. This approach has limited effectiveness because:

• You cannot control how AI systems link to your products
• AI-generated links typically don't preserve your UTM parameters
• The approach only works for direct click-through traffic
• Implementation requires AI platforms to cooperate with your tracking

Survey-Based Attribution

Post-purchase surveys asking "how did you hear about us?" can capture some AI influence that GA4 misses. However:

• Survey response rates are typically low
• User recall is imperfect and often incorrect
• AI influence may be subconscious and unreported
• Surveys add friction to the purchase experience

Correlation Analysis

Some teams attempt to correlate traffic and revenue patterns with known AI activity—product mentions, ChatGPT feature launches, etc. This approach can surface suggestive patterns but cannot establish causation and struggles to quantify AI's actual contribution.

Brand Search Proxies

Increases in branded search traffic after AI exposure might proxy for AI influence. But brand search is affected by many factors, and isolating the AI component is analytically challenging without additional data sources.

All these workarounds share a common limitation: they attempt to infer AI influence indirectly rather than measuring it directly. They can supplement understanding but cannot substitute for purpose-built AI commerce measurement.

Complementing GA4 for AI Commerce

The solution isn't abandoning Google Analytics—it remains valuable for understanding owned property behavior and traditional digital marketing performance. The solution is complementing GA4 with measurement approaches designed specifically for AI commerce.

What Purpose-Built AI Measurement Provides

Platforms designed for AI commerce fill the specific gaps GA4 cannot address:

AI surface monitoring: Direct observation of how AI systems present your products, what recommendations they make, and how your visibility changes over time.

Product data interpretation: Understanding how AI systems parse and utilize your product information—revealing optimization opportunities invisible to GA4.

Competitive visibility: Comparative analysis showing your AI presence relative to competitors, identifying gaps and opportunities.

AI-specific attribution: Methodologies designed for the unique attribution challenges of AI commerce, moving beyond click-based models.

Predictive signals: Leading indicators of AI visibility changes that can predict revenue impact before it manifests in traditional metrics.

The Complementary Approach

Leading brands are implementing layered measurement strategies that combine:

• GA4 for owned property analytics and traditional digital marketing measurement
• AI commerce platforms for AI visibility monitoring and AI-specific attribution
• Survey instruments for qualitative validation of AI influence
• Correlation frameworks for connecting AI signals to business outcomes

This layered approach recognizes that no single platform can provide complete AI commerce measurement—but the combination of purpose-built tools can approximate comprehensive visibility.

Building Organizational Capability

Beyond tools, addressing the AI commerce measurement gap requires organizational changes:

Metric evolution: Expanding KPI frameworks beyond GA4-centric metrics to include AI visibility, AI share of voice, and AI-attributed revenue estimates.

Reporting integration: Building dashboards that synthesize GA4 data with AI commerce platform data to provide holistic performance views.

Analytical skills: Developing or hiring analytical capabilities specific to AI commerce—understanding AI systems, interpreting AI visibility data, and constructing attribution models suited to AI influence patterns.

Cross-functional alignment: Ensuring marketing, e-commerce, and product teams share understanding of AI commerce measurement limitations and work collaboratively on solutions.

The Path Forward

Google Analytics 4 isn't broken—it was designed for different problems. But relying solely on GA4 for AI commerce measurement is like using a hammer to drive screws. The tool works excellently for its intended purpose but fails at a fundamentally different task.

The brands that will thrive in AI commerce are those that accept GA4's limitations and invest in complementary measurement approaches. This requires:

1. Acknowledging the blind spot: Stop assuming GA4 tells the complete story of AI influence
2. Investing in AI-specific measurement: Adopt platforms designed for AI commerce visibility and attribution
3. Building integrated analytics: Create frameworks that synthesize traditional and AI-specific data
4. Developing new capabilities: Build organizational skills suited to AI commerce measurement

The cost of inaction is continuing to fly blind as AI captures an increasing share of product discovery. The brands that master AI commerce measurement will understand their true competitive position—and will optimize effectively while competitors remain in the dark.

---

Related Reading:

• The AI Commerce Tool Landscape: Understanding Your Options
• Why Your Marketing Team Can't See the AI Problem
• Why You Can't Track AI-Influenced Purchases
• Feed Management Tools vs. AI Commerce Platforms
• AI Commerce Tools Compared: Finding the Right Solution

---

Your analytics platform has a blind spot. Ready to see what GA4 can't show you about AI commerce?

---

Want to see how your store scores? Run a free AI readiness scan and get your store's AI visibility report in 60 seconds.

---

About the Author: Josh is the founder of Noema, an AI commerce observability platform that helps e-commerce brands understand how AI shopping agents see their products. Noema has scanned 80,000+ Shopify stores to build the industry's most comprehensive AI readiness benchmarks.