How to Track Traffic from AI Overviews in Google Analytics 4 – Implementation Guide

Listen to the audio version of this article.

There are two challenges for analysts

1. Identifying AI Referrals (Referring Traffic): Separating traffic originating from LLMs (in the form of referrals) from standard referring traffic. This topic is widely discussed, but we include it in our instructions to provide a comprehensive solution.

2. Identifying AI Overviews Traffic: Understanding what portion of the traffic previously labeled as Organic Search actually comes from AI Overviews. This is the first guide where you will find instructions on how to measure this precisely.

Your Advantage: Precision and User Intent

The unique value of this Guide lies in integrating these methods through a modular modification of data in Google Tag Manager (GTM):

• Modification of utm_medium (Our Know-How): This is the key element: at the Tag Manager level, we modify the medium attribute for hits coming from AI Overviews, overriding it to the unique value "ai-overviews". This allows us to create a consistent AI Search channel group in GA4 that works more precisely than standard grouping.

• Insight into Intent (Optional, but Crucial for SEO): We append the attributes snippet_text_start and snippet_text_end to every click from AI Referrals. This allows you to see exactly what the user was looking for, providing invaluable insight into query intent.

Below, we present our proprietary solution that will allow you to identify this traffic and, most importantly, determine which exact fragment of your text was used by the AI.

Step 1: Import the GTM Container

1. Download the prepared JSON file (ai overview tracking - GTM template by beecommerce.pl.json).

2. Log in to Google Tag Manager and go to the Admin section.

3. Select Import Container.

4. Select the downloaded JSON file and set the import method to Merge. 

5. Choose the option Overwrite conflicting tags, triggers and variables. 

6. Click Add to workspace.

Step 2: GA4 Tag Configuration

You need to attach the imported variables to your main GA4 configuration tag so that all events automatically include information about AI Overviews.

1. Open your existing GA4 Config Tag (GA4 Configuration Tag).

2. Expand the Shared Event settings section. If you have already configured this variable, proceed to point 5 of the instructions.

3. From the Event settings variable list, select New Variable... (New Variable).

4. Name the new variable "GA4 | Var | Event Settings | Common Event Settings".

5. Add the following three parameters:

• Event parameter: medium | Value: {{Google Snippet Tracking | Var | JS | Replacement Medium}}

• Event parameter: snippet_text_start | Value: {{Google Snippet Tracking | Var | JS | Snippet Start}}

• Event parameter: snippet_text_end | Value: {{Google Snippet Tracking | Var | JS | Snippet End}}

Hint: Start typing {{ in the Value field to use the variable search function and avoid typos.

6. Save the variable, and then save the GA4 configuration tag.

Step 4: Creating Custom Dimensions

These dimensions will allow you to analyze which text fragment from your page was used by the LLM and generated traffic in the case of AI Referrals.

1. Log in to GA4 and go to Admin > Data display > Custom definitions.

2. Click Create custom dimension.

3. Create two dimensions with Event scope:

• Dimension 1:

  • Dimension name: Snippet Start

  • Event parameter: snippet_text_start (Enter the name manually if it's not on the list).

• Dimension 2:

  • Dimension name: Snippet End

  • Event parameter: snippet_text_end (Enter the name manually).

4. Save the changes.

How to create a new traffic channel in GA4?

In GA4, in the Admin section, go to Channel groups and click Create new channel group.

2. Name this channel grouping "Default channel group with AI Search".

3. Set the new Channel Grouping as Default.

7. Add an OR rule: Medium | matches exactly ($=$) | ai-overviews

8. Save the channel.

9. Move the channel ABOVE the Referral line. This is absolutely crucial because otherwise, the Referral override rule will never trigger, and you won't see the data in the "AI Overviews" group.

10. Done.

What to Expect in Default Reports? (New, True Data)

Implementing Our solution will radically change the traffic picture in the existing channels. Be ready for these changes, as they signify the correct attribution of data to the new AI Search channel:

• Referrals Drop: You will notice a 50-80% drop in sessions in the Referral channel.

• Organic Search Drop: Sessions in the Organic Search channel may drop by 30-50%.

Important: You are not losing data! You have simply isolated the previously masked sessions into a dedicated AI Search channel. To see the detailed source of AI traffic (e.g., ChatGPT, Claude), use the segmentation by utm_source (Source) in the default reports.

Depth of Analysis: Using Explorations

If you need deeper analysis and want to examine the context used in the AI, leverage the custom dimensions Snippet Start and Snippet End in Explorations.

How to create a contextual report:

1. Log in to GA4 and go to the EXPLORATIONS section.

2. Create a new report based on Free-form Exploration.

3. Add as Dimensions: Snippet Start, Snippet End, and Landing page + query string.

4. Add as Metrics: Sessions.

5. Apply a filter to display only sessions where the Default Channel Group is AI Search.

Thanks to this report, you will see not only the number of sessions but also exactly which fragment of text (starting at Snippet Start and ending at Snippet End) attracted the user to the page.