Outline Technologies — SEO, AEO & GEO Agency
Back to Blog
GEO AnalyticsComplete Guide

How to Track Your Brand's AI Citations in 2026 (ChatGPT, Perplexity, Gemini)

70.6% of AI referral traffic shows up as Direct in Google Analytics. Your brand could be appearing in thousands of ChatGPT and Perplexity answers without you knowing. Here's how to actually measure it.

Ahmed Shanti 14 min readMay 13, 2026
In This Guide
The dark traffic problemGA4 custom channel groupsUTM parameters for AI trafficPlatform-specific trackingAI SOV: the key metricDedicated monitoring toolsManual brand mention auditsBuilding your tracking dashboardBenchmarks and what good looks like

The Dark Traffic Problem

In 2022, "dark traffic" — website visits with no referrer data — was a minor analytics nuisance. In 2026, it's a measurement crisis. SparkToro and Datos research found that 70.6% of AI referral traffic arrives as Direct in Google Analytics 4. When someone asks ChatGPT about your brand, gets a response that mentions your site, and clicks through — GA4 sees that as Direct traffic. No source. No medium. No way to know it came from an AI.

The problem compounds because different AI platforms handle referrers differently. Perplexity passes referrer data fairly consistently (you'll see perplexity.ai in your referrals). ChatGPT links from within the app strip the referrer entirely. Gemini varies by surface. Claude usually strips it. The result is a fragmented picture that makes AI-sourced traffic nearly invisible in standard analytics setups.

Why This Matters More Than Click Volume

Even when users don't click, AI citations build brand awareness and credibility. Tracking clicks is only half the picture. You need to measure how often your brand appears in AI-generated answers — what practitioners call AI Share of Voice (AI SOV) — separately from how many of those appearances turn into website visits.

Setting Up GA4 Custom Channel Groups for AI Traffic

The first step is capturing every referral you cansee. GA4's default channel grouping lumps many AI referrers into "Unassigned." A custom channel group fixes that.

In GA4, go to Admin → Data Settings → Channel Groups → Create new channel group. Add a channel called "AI Search" and define it with the following source/medium conditions:

GA4 AI Channel Group Regex Pattern

Source matches regex:
perplexity\.ai|chat\.openai\.com|chatgpt\.com|
gemini\.google\.com|bard\.google\.com|
copilot\.microsoft\.com|bing\.com\/chat|
claude\.ai|you\.com|phind\.com|
poe\.com|character\.ai

Place this channel group above "Direct" and "Unassigned" in the priority order. GA4 assigns sessions to the first matching rule.

Once configured, you'll see an "AI Search" channel in your acquisition reports. It won't capture the dark traffic, but it will correctly attribute the clicks that do pass referrer data — primarily Perplexity and some Copilot traffic.

UTM Parameters: Your Best Defense Against Dark Traffic

For traffic you control — links in your own AI-visible content — UTM parameters let you tag and trace even when referrers are stripped. The key is being consistent.

Tag your primary pages. Add utm_source=ai-answer&utm_medium=generative&utm_campaign=geo to links in your highest-priority content. When Perplexity or ChatGPT cites and users click, you'll see these UTMs in GA4 even without a referrer.
Create platform-specific variants. Use utm_source=perplexity, utm_source=chatgpt, utm_source=gemini where you can confirm which platform is citing you (e.g., in your llms-full.txt links or verified content syndication).
Tag your featured snippet targets. Pages optimized for Google's AI Overviews should carry utm_campaign=ai-overviews so you can separate this traffic from standard organic.
Build a UTM taxonomy document. Standardize all your AI-related UTM values in a shared doc. Inconsistent tagging produces unusable data — 'AI_Answer' and 'ai-answer' are different segments in GA4.

Platform-Specific Tracking Approaches

Perplexity

Perplexity is the most transparent AI platform for tracking. Their crawler (PerplexityBot) hits your pages regularly, and referral traffic from perplexity.ai shows up in GA4 with referrer intact. Monitor your Referrals report for perplexity.aiand create a dedicated segment for it. You can also sign up for Perplexity's Publisher Program, which gives you verified citation data and some revenue sharing on clicks.

ChatGPT

ChatGPT is harder. Links within ChatGPT conversations strip referrers, so most ChatGPT-driven traffic appears as Direct. The most reliable signal is a spike in branded search + direct traffic together after a period when your brand appears in popular ChatGPT outputs. You can also use tools like Profound or Semrush's AI product to directly query ChatGPT with your brand-relevant questions and see whether you're cited.

Google Gemini / AI Overviews

Google Search Console now labels some AI Overview traffic separately in its reports. Monitor your GSC performance data filtered by "AI Overviews" appearance type. For Gemini.google.com as a direct interface, it functions similarly to other AI chat — referrer data is inconsistent. The bigger opportunity is Google's AI Overviews in search results, which you can track through GSC.

AI Share of Voice: The Metric That Actually Matters

Clicks and referrals only tell you about the users who clicked. AI SOV tells you how often you appear in AI-generated answers regardless of click behavior. This is the metric that correlates most strongly with brand consideration and long-term organic traffic growth.

How to Calculate AI SOV

AI SOV = (Queries where your brand appears ÷ Total queries tested) × 100

Run 50–100 of your most important keyword queries through ChatGPT, Perplexity, and Gemini. Record how many times your brand is mentioned or cited versus competitors. Calculate the percentage for each platform separately — your Perplexity SOV will likely differ significantly from your ChatGPT SOV.

Track AI SOV weekly against competitors. If your SOV drops across all platforms simultaneously, it usually means a competitor has published stronger, more-cited content in that topic area. If it drops on one platform only, there may be a technical issue (e.g., your robots.txt blocking that platform's crawler).

Dedicated AI Monitoring Tools

Manual tracking is a start, but dedicated tools automate the heavy lifting. Here are the main options as of 2026:

Profound. The most purpose-built AI visibility platform. Tracks brand mentions across ChatGPT, Perplexity, and Gemini at scale. Provides sentiment analysis on how your brand is described, not just whether it appears. Enterprise-priced but the most comprehensive.
Semrush AI Toolkit. Semrush added AI brand tracking features in 2025. If you're already a Semrush user, this is the lowest-friction option. Queries AI platforms for your target keywords and tracks citation frequency over time.
Otterly.AI. Specializes in AI visibility monitoring with particular depth on Perplexity. Provides citation rate data, competitor comparisons, and integrates with GA4 to correlate AI mentions with traffic patterns.
BrandMentions + manual AI queries. The budget-friendly option. Use BrandMentions for web-based brand monitoring, supplement with a structured weekly manual query process across AI platforms using a spreadsheet tracker.
Outline's citation tracking. We built citation tracking natively into Outline's dashboard — real-time AI mention monitoring across the major platforms with trend data and competitive benchmarking.

Building Your AI Citation Tracking Dashboard

Whether you use a dedicated tool or build your own, your tracking dashboard should answer five questions every week:

1

How often is my brand mentioned across ChatGPT, Perplexity, and Gemini for my target queries? (AI SOV)

2

Is that rate trending up, down, or flat compared to last week and last month?

3

Which competitors are mentioned more frequently than we are, and in what contexts?

4

What traffic is arriving from AI referrals with confirmed source attribution (Perplexity, Copilot)?

5

How much Direct traffic correlates with AI mention spikes — a proxy for unattributed AI clicks?

Benchmarks: What Good Actually Looks Like

Without context, AI SOV numbers are meaningless. Here are rough benchmarks based on aggregated data from B2B SaaS and marketing companies as of mid-2026:

StageAI SOV RangeWhat It Means
Early / No GEO work0–5%Brand rarely appears in AI answers for target queries
Building visibility5–20%Appearing in niche/specific queries, not broad ones
Competitive20–40%Regular appearances; some competitor parity
Category leader40–60%Consistent top-of-mind presence in AI answers
Dominant60%+Brand is the default recommendation for its category

Most brands starting AI tracking today will find themselves in the 0–5% range. The brands that started GEO work in 2024–2025 are now in the 20–40% range. The compounding effect is real — the longer you've been building authoritative, citable content, the higher your AI SOV climbs.

The Correlation You Should Watch

Brands with high AI SOV consistently see their branded organic search volume grow — people who encounter your brand in ChatGPT often then Google you directly. Track both metrics together. If your AI SOV is climbing but branded search is flat, it means awareness is building but brand recall is weak. That's an entirely different problem to solve.

Quick-Start: Your First AI Tracking Week

Set up GA4 custom channel group for AI referrers (30 min)
Run 50 target queries manually across ChatGPT, Perplexity, Gemini — record citation results in a spreadsheet
Calculate baseline AI SOV for each platform
Identify top 3 competitors appearing more frequently than you
Note the exact contexts/queries where you appear vs. don't appear
Set calendar reminder to repeat this process weekly
Evaluate one dedicated tracking tool (Profound free trial or Semrush AI Toolkit)
Configure GA4 to flag anomalies in Direct traffic (possible AI citation spikes)
AS
Written by Ahmed Shanti
Co-Founder, Outline Technologies

Ahmed built Outline's AI citation tracking system after spending months wrestling with the dark traffic problem firsthand. He writes about measurement, attribution, and the evolving analytics stack for AI-era marketing.