---
title: "How to Measure Local AI Referral Traffic in GA4: A Practical Guide"
url: "https://jayyseo.com/measure-local-ai-referral-traffic-ga4/"
date: "2026-11-13T10:37:25+00:00"
updated: "2026-11-13T11:30:28+00:00"
author: "Jay"
description: "Learn how to measure local AI referral traffic in GA4 with 7 steps: date ranges, the AI Assistant channel, regex, city views, key events and attribution gaps."
category: "AEO"
---

# How to Measure Local AI Referral Traffic in GA4: A Practical Guide

Local AI referral traffic is the set of website visits that come from links inside AI answers, such as ChatGPT, Gemini, Perplexity, Claude and Microsoft Copilot, and land on the pages for your local business. GA4 can show you how many of these visits you get, where they land and whether they turn into calls, forms or bookings.

Picture a family dental clinic in Austin. The owner hears that patients now ask ChatGPT for “the best family dentist in Austin” and wants to know if any of those people reach the website. GA4 answers most of that question once the setup is right.

I use RadarKit alongside GA4 for this work. RadarKit shows whether AI recommends the business in the first place, and GA4 shows what happens after someone clicks. This guide covers the GA4 side in seven steps, then shows where RadarKit fits.

## What Is Local AI Referral Traffic?

A referral is a visit that arrives from another website or app. When the referring site is an AI assistant, GA4 can now label it on its own. Since May 13, 2026, GA4 has an AI Assistant channel in the Default Channel Group, and it sets the medium to “ai-assistant” when the referrer matches Google’s list of assistants.

The local part of local AI referral traffic is up to you. For a single-location business, it usually means AI visits that land on your service pages. For a multi-location brand, it means AI visits that land on a specific location page, like /austin/ or /dallas/.

A useful report answers three questions:

- How many visits did AI assistants send to this location in the date range?
- Which pages did those visitors land on?
- How many of those visits turned into a key event, like a booking or a call?

## How to Measure Local AI Referral Traffic in GA4: 7 Steps

Here is the process I follow to measure local AI referral traffic.

## Step 1: Set Your Traffic or Signup Question and Date Range

Start with one question. A traffic question sounds like “How many AI assistant sessions reached our Austin page last month?” A signup question sounds like “How many booking forms came from AI assistant visits in the last 90 days?”

Next, choose the date range. Local AI referral traffic to one location is often a small number, so a full month or the last 90 days gives a steadier read than a single week. If you already set a [local AI visibility baseline](https://jayyseo.com/how-to-set-a-local-ai-visibility-baseline-5-steps/), use the same dates so the two reports line up.

Before you build anything, check data retention under Admin, Data collection and modification, Data retention. Explorations only reach back as far as this setting allows, and the default is two months. Raise it to 14 months if you want year-over-year views later.

## Step 2: Check the AI Assistant Channel in Traffic Acquisition

The fastest first look at local AI referral traffic is the built-in channel. Open Reports, Acquisition, Traffic acquisition, set the primary dimension to Session default channel group, and find the AI Assistant row.

Now add Session source as a secondary dimension. This shows which assistants sent the visits. Google’s channel definition names ChatGPT, Gemini, DeepSeek, Copilot and Grok as examples, and the launch note names Claude too.

Google has not published the full list of recognized assistants. If you see an assistant like Perplexity sitting in the Referral row, that is normal, and Step 3 fixes it.

## Step 3: Build a Custom Channel Group With a Regex Filter

A custom channel group lets you choose exactly which assistants count toward your local AI referral traffic. It also applies to past data, which is handy for the months before May 2026.

Go to Admin, Data display, Channel groups, and click Create new channel group. GA4 starts you with a copy of the default group. Then add a channel named “AI assistants” with two conditions joined by OR:

1. Medium exactly matches **ai-assistant**
2. Source matches regex:

`.*(chatgpt\.com|chat\.openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai).*`

Click Reorder and drag “AI assistants” above Referral, then save the group. GA4 places each session in the first channel whose rule it matches, so the order matters.

A few notes from setting this up:

- Standard properties allow two custom channel groups, so name this one clearly.
- You need the Editor role or higher to create a group.

## Step 4: Add a Location View With City and Landing Page

This is where your report turns into true local AI referral traffic data. Open Explore and create a free-form exploration.

| Setting | What to add |
| --- | --- |
| Rows | Your custom channel group, then Landing page |
| Columns | City |
| Values | Sessions, Engaged sessions, Key events |
| Filter | Custom channel group exactly matches “AI assistants” |

The City dimension tells you where the visitor was, based on their IP address. The landing page tells you which location they chose. For a multi-location business, I trust the landing page more, because it reflects the location the visitor actually picked.

If you manage many locations, this table plugs straight into a [multi-location AI report](https://jayyseo.com/how-to-build-a-multi-location-ai-report-5-steps/). Keep the same filter on every location so the numbers compare fairly.

## Step 5: Set Up Events and Mark Your Key Events

Local AI referral traffic matters most when it turns into customers, so the next step is tracking the actions that count.

For a local business, three actions matter most:

- **Booking or contact form:** GA4’s recommended event name for this is generate_lead.
- **Phone click:** a click on a tel: link, which usually needs a small tag in Google Tag Manager.
- **Directions click:** a click on your map or directions link.

Once these events start firing, mark them as key events under Admin, Data display, Key events. Key events is GA4’s current name for what used to be called conversions.

## Step 6: Run Conversion Checks in an Exploration

Go back to your exploration from Step 4. Add Key events and Session key event rate as values, then break down by Event name.

Now you can answer your signup question about local AI referral traffic directly. For example, you can see how many generate_lead events came from AI assistant sessions that landed on the Austin page in the last 90 days.

I like to compare local AI referral traffic against Organic Search for the same landing pages. AI visitors may arrive with part of their question already answered, so their engagement rate is worth watching side by side.

## Step 7: Explain Missing Attribution and What the Data Can Prove

GA4 counts the local AI referral traffic it can identify. That gives you a clear, honest number, and it is worth knowing where that number stops.

| What happens | Where it shows in GA4 |
| --- | --- |
| Click from ChatGPT, Gemini or Claude with a referrer | AI Assistant channel |
| Click from Google AI Overviews or AI Mode | Organic Search |
| Link copied from a chat and pasted into a browser | Direct |
| Click from some apps that pass no referrer | Direct |
| ChatGPT link that carries utm_source=chatgpt.com | Source shows chatgpt.com, caught by your regex |
| Someone reads the answer and calls you without clicking | Not in GA4 |

So the data can prove three things: a minimum count of AI visits, the pages they landed on, and the key events those visits produced. It cannot show how often AI recommended you without a click. Write “minimum” next to your local AI referral traffic number, and readers will read it the right way.

Google AI Overviews and AI Mode deserve one extra note. Because their clicks count as Organic Search, you can watch them in Search Console. My guide on [using Google Search Console for local AI prompts](https://jayyseo.com/how-to-use-google-search-console-for-local-ai-prompts-5-steps/) covers that side.

## How RadarKit Completes the Picture

GA4 measures local AI referral traffic after the click. RadarKit measures the moment before it, when an AI assistant decides which businesses to recommend.

I set up the prompts a local customer would ask, like “best family dentist in Austin,” in RadarKit’s Local Radar. RadarKit runs them in the real ChatGPT, Gemini, Perplexity, Microsoft Copilot, Google AI Mode and Google AI Overviews interfaces, from residential IPs. For each answer, it records whether the business appears, its position and every competing business the answer names. You can read more in my [RadarKit Local Radar review](https://jayyseo.com/radarkit-local-radar-review/).

Here is how the two tools work together on local AI referral traffic:

| Question | GA4 | RadarKit |
| --- | --- | --- |
| Did AI recommend us? | No | Yes, with appearance rate and position |
| Which sources did the answer cite? | No | Yes, cited and consulted sources |
| Did people click through to our site? | Yes | Shown in its AI Traffic Analytics view, which reads from GA4 |
| Did those visits turn into bookings? | Yes, with key events | Read this in GA4 key events |
| AI Overviews and AI Mode visibility | Mixed into Organic Search | Tracked as separate engines |

When RadarKit shows which pages an answer cites, you can see which of your pages earn clicks. My guide on [finding the pages cited for local searches](https://jayyseo.com/how-to-find-the-pages-cited-for-local-searches-5-step/) walks through that process.

![RadarKit Local Radar next to local AI referral traffic data from GA4](https://jayyseo.com/wp-content/uploads/2026/11/local-ai-referral-traffic-radarkit-local-radar.jpg) ![RadarKit overview of local AI referral traffic visibility](https://jayyseo.com/wp-content/uploads/2026/11/local-ai-referral-traffic-radarkit-overview.jpg)

### RadarKit Key Features

RadarKit is the tool I use for the visibility side of this work, before local AI referral traffic shows up in GA4.

- **AI engine tracking:** RadarKit asks your prompts in the real ChatGPT, Google AI Mode, Google AI Overviews, Gemini, Perplexity and Microsoft Copilot interfaces, from residential IPs in 50+ countries, so you see the answers a customer would see before you measure local AI referral traffic.
- **Prompts and topics:** You add the questions your customers ask, group them into topics, and RadarKit runs them on a schedule, or you can run a prompt right away for a fresh check.
- **Visibility Score:** This score shows the share of tracked prompts and models that mention your brand, with a trend against the previous period, so you can see how often you are recommended before local AI referral traffic arrives.
- **Brand Reputation:** This score tracks how AI talks about your brand over time.
- **Average Position:** When your brand appears, RadarKit records where it sits in the recommendation list and how that compares with other domains.
- **Sentiment:** RadarKit labels mentions as positive, mixed or negative, overall and per model, and surfaces repeated phrases about your brand.
- **Local Radar:** Local Radar shows whether AI recommends your business, your appearance rate, your position, and every competing business the answer names, with category, rating and review count. You can filter by listing platform, such as Yelp or Google Business Profile, and see which prompts each business wins. Self-serve plans cover 50+ countries, and state and city targeting is part of Enterprise. This is the step before local AI referral traffic reaches your site.
- **Shopping Radar:** Shopping Radar records the product, store, price, rating and position in AI product carousels, plus how often your products appear and which stores win.
- **Ads Radar:** Ads Radar captures the ads shown next to AI answers, including the advertiser, headline, description, landing page, creative, and how often those ads appear.
- **Citation Data:** Citation Data shows the pages AI cites or only consults, which of those pages link back to your site, and the content formats cited most, so you can see which pages may send local AI referral traffic.
- **Page Radar:** Page Radar lets you watch a specific page, including a guest post, and see which models cite it and on which prompts.
- **Query Fanout:** Query Fanout records the web searches a model ran before it answered, and how often models search at all.
- **Competitors and Share of Voice:** RadarKit lists the brands and domains recommended in the same answers, with share of voice, average position and rank.
- **Create Content:** The Content Agent studies the pages AI cites for a prompt, plus Google results, query fanout and People Also Ask questions, then drafts an article in your brand voice, including blog posts, listicles, comparisons, alternatives, FAQs and first-person reviews.
- **Optimize Content:** You can send an existing article to be reworked around the entities and query fanouts that AI already cites, then edit it and publish to WordPress.
- **AI Traffic Analytics:** RadarKit connects to Google Analytics 4 with read-only access so you can view AI traffic next to your visibility data and connect recommendations to local AI referral traffic.
- **Plan Agent and Google Search Console:** Every plan can connect Google Search Console. The Plan Agent reads that data with your RadarKit citations, finds gaps, and builds a task list for the other agents.
- **Outreach Agent:** The Outreach Agent finds sites that cite your competitors, drafts on-brand emails, and can send them through your Gmail after you approve, then follow up.
- **Reddit Agent:** The Reddit Agent opens the Reddit threads AI cites in your space, shows where your brand is missing, and surfaces the discussion.
- **Chat Agent:** The Chat Agent answers questions about your project in plain language, using your live RadarKit data.
- **API and MCP access:** Plans include a REST API and an MCP server, so you can pull the same data into ChatGPT, Claude, Cursor or Codex.
- **Exports, branded PDFs and the Client Portal:** You can export responses, sources, competitors, rankings, query fanout and radar data as a file, produce branded PDF reports (2 on Lite, 10 on Growth and 20 on Pro each month), and share a live, view-only Client Portal or a read-only report link.
- **Full history:** Every plan keeps full historical data, and the dashboard lets you compare a date range with the previous period, so you can watch visibility next to local AI referral traffic.
- **Projects and workspaces:** You can run more than one project if you manage several locations or brands: 2 on Lite, 5 on Growth, 10 on Pro, and unlimited on Enterprise.
- **Project settings and team seats:** Project settings hold your brand kit, brand and competitor aliases, timezone and roles, with 2 extra seats on Growth, 5 on Pro, and unlimited seats plus single sign-on on Enterprise.
- **Free GEO tools:** RadarKit also offers free tools with no signup, including an LLMs.txt checker, an LLMs.txt generator and an AI SEO Chrome extension.

## RadarKit Pricing

RadarKit tracks six AI platforms on every plan.

| Plan | Price | Prompts tracked | Refresh |
| --- | --- | --- | --- |
| Lite | $29 per month | 15 | Every 72 hours |
| Growth | $79 per month | 50 | Daily |
| Pro | $139 per month | 100 | Daily |
| Enterprise | Custom | Custom | Daily or weekly |

Self-serve plans track local recommendations from residential IPs in 50+ countries. State and city level targeting is part of Enterprise. Every self-serve plan includes a 7-day free trial and the Google Analytics connection.

## Final Thoughts

Measuring local AI referral traffic in GA4 is a seven-step habit. Set one question and a date range, check the AI Assistant channel, build a custom channel group, add a location view, set key events, run conversion checks and explain what the data can prove.

GA4 tells you what happened after the click. RadarKit tells you whether AI recommended you before it. Put both side by side each month, and you will see how your local AI referral traffic grows as your visibility grows.

## FAQ

### 1. What counts as AI referral traffic in GA4?

It’s any visit that comes from a link inside an AI tool’s answer, such as ChatGPT, Perplexity, Gemini or Copilot. GA4 logs these visits as referrals, and you can spot them by their source domain.

### 2. Why doesn’t GA4 show every visit that comes from AI tools?

Some AI apps don’t pass referrer data, so those visits show up as direct traffic. Many people also read the AI’s answer, then search for your business name or call you without ever clicking a link.

### 3. How can I see which cities my AI visitors come from?

Filter your AI sources in an Exploration report, then add City as a dimension. This tells you whether AI tools are sending people from your service area or from places you don’t serve.

### 4. How does RadarKit help when GA4 only shows clicks?

GA4 shows you the people who clicked. RadarKit shows how often your business shows up in AI answers in the first place. Put the two side by side and you can tell whether a drop in traffic means fewer mentions or fewer clicks.

### 5. Can RadarKit track local AI visibility for different service areas?

Yes. RadarKit’s Local Radar shows how your business appears in AI answers for local searches. You can match those results against the city data in GA4 to see which areas need more work.

---

**Is AI recommending your brand?** Track your brand in Google AI Overview, AI Mode, ChatGPT, Perplexity and more with RadarKit. No payment upfront. Cancel anytime.

[Start your 7-day free trial](https://radarkit.ai/register)

---

**About the author:** I am Sanjay Singh (Jay), co-founder of Radarkit.ai, AI-search visibility software that helps businesses understand and improve how they appear in LLMs like ChatGPT, Perplexity, Gemini, Claude and Microsoft Copilot.

---

Source: https://jayyseo.com/measure-local-ai-referral-traffic-ga4/
