iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Short answer: For most teams, Google Analytics 4 (GA4) is the easiest starting point because its May 13, 2026 update added an AI Assistant channel. Choose Matomo when you need separate human-referral and crawler reports, Plausible or Fathom for privacy-focused analytics with low implementation overhead, and Adobe Analytics when your enterprise already depends on Analysis Workspace and configurable dimensions. None of these tools can count every AI mention; they measure observable visits or requests that reach your site.
Which analytics tool is best for AI-assistant referrals?
The right choice depends on what you need to know after someone clicks a link in ChatGPT, Gemini, Claude, Perplexity, Copilot, or another assistant.
![Google Workspace Guide: Unlock Every Google App – Elevate Efficiency with Exclusive Tips, Time-Savers & Step-by-Step Screenshots for Quick Mastery [II EDITION]](https://m.media-amazon.com/images/I/41CX0AjJhAL._SL160_.jpg)
| Tool | Best fit | AI-referral capability | Machine-traffic handling | Outcome reporting |
|---|---|---|---|---|
| GA4 | Organizations already standardized on Google Analytics | Dedicated AI Assistant channel; custom regex channel groups | Primarily classifies observed referral traffic | Google Analytics conversions, audiences and revenue reports |
| Matomo | Teams needing privacy and a clear human-versus-automation split | AI Assistants acquisition channel in Cloud and On-Premise 5.5.0+ | Separate AI Agent and AI Chatbot reports | Visits, goals, row evolution and segmented visit logs |
| Adobe Analytics | Large enterprises using Analysis Workspace | Conversational AI tools referrer type | Documents scraping, raw-HTML retrieval and agentic workflows separately | Highly configurable dimensions and Workspace analysis |
| Plausible | Simple, privacy-focused reporting | Automatic AI Assistants channel for recognized tools | Known AI crawlers and bots are filtered | Landing pages, engagement, goals, funnels, signups, orders, revenue and average order value |
| Fathom | Lightweight analytics with export and API needs | AI bucket with per-tool filtering | Known bots are filtered; unidentifiable traffic can be Direct | Historical reports, CSV export and an ai_source API dimension |
There is no authoritative cross-vendor accuracy benchmark or universal percentage for AI-origin traffic. Vendor recognition lists, browser privacy settings, referrer behavior and your tagging determine what appears in reports.
Recommended Free Tools
What “AI-assistant referral” actually measures
An AI-assistant referral is an observable request or page visit that arrives after a user clicks a link presented in an assistant response. The analytics system usually identifies it from an HTTP referrer, a campaign parameter, or both. It is not a count of every time a model mentions your company.
- Human click-through: A person reads an answer, clicks your link and loads a page where the analytics script runs.
- Automated retrieval: An assistant, crawler or agent requests content directly. It may not execute JavaScript, so browser analytics can miss it.
- Unattributed arrival: Referrer information can be stripped by an app, browser or privacy control. A copied link or an untagged URL can appear as Direct.
Adobe describes three distinct AI interactions: pretraining scrapes, on-demand prompts that may fetch raw HTML without running JavaScript, and agentic workflows. Only the latter two can produce site hits, and an on-demand HTML fetch may never become a browser analytics session. Matomo likewise separates visitors referred by assistants from automated systems that retrieve or request content.
GA4: the best starting point for teams already using Google Analytics
Google Analytics Help announced on May 13, 2026 that GA4 can measure recognized assistant referrals with a dedicated channel. The classification uses medium ai-assistant, channel AI Assistant and campaign (ai-assistant). Google describes this as a dedicated way to measure and analyze traffic originating from popular AI assistants.
What the dedicated channel gives you
Once a recognized referrer is classified, you can use AI Assistant as a channel in acquisition reports and compare users, sessions, engagement, conversions and revenue with Organic Search, Referral, Email and other channels. Existing GA4 conversion events and ecommerce values continue to apply, so the useful question is not only “How many visits?” but also “Which assistant visitors complete a valuable event?”
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →How to track ChatGPT traffic in GA4
- Open your GA4 property and go to the acquisition reports. Select the channel or session-default-channel-group dimension.
- Filter for AI Assistant. Add landing page, assistant source, engaged sessions and your conversion events as secondary dimensions or report columns.
- For a conversion-focused view, use the same date range and attribution model you use for other channels, then compare key events, purchase revenue or lead value.
- Break the channel down by source when the property exposes the individual referrer. This lets you distinguish, for example, ChatGPT from Gemini or Perplexity rather than treating all assistants as one row.
Adding assistants that GA4 does not recognize
GA4 also supports custom channel groups with regular-expression matches for assistant URLs. Create a custom channel, add the relevant hostnames or URL patterns, and put the custom AI channel above Referrals in the channel-group order. Channel rules are evaluated in order; leaving AI below Referrals can cause the same visit to be classified as a generic referral first.
Maintain the expression as assistants add domains, change redirect behavior or move from one hostname to another. Keep a change log so a trend break can be explained when a rule changes. A custom rule cannot recover a referrer that was never sent, and it cannot prove that an automated request represents a person.
Rank #2
![The Google Workspace Bible: [14 in 1] The Ultimate All-in-One Guide from Beginner to Advanced | Including Gmail, Drive, Docs, Sheets, and Every Other App from the Suite](https://m.media-amazon.com/images/I/41AWmhmGFuL._SL160_.jpg)
- The Google Workspace Bible: [14 in 1] The Ultimate All in One Guide from Beginner to Advanced Including Gmail, Drive, Docs, Sheets, and Every Other App from the Suite
- ABIS BOOK
GA4 limitations
- GA4 primarily classifies referral traffic that its collection tag observes; it does not provide Matomo-style dedicated AI-agent and AI-chatbot reports.
- Referrer stripping, copied links and app webviews can move legitimate assistant visits into Direct.
- A request that does not execute the GA4 script will not appear as a normal session.
- Custom regex lists require ongoing maintenance, even when the built-in channel handles the major assistants.
Matomo: the clearest human-versus-machine model
Matomo Cloud provides an AI Assistants acquisition channel covering ChatGPT, Copilot, Gemini, Claude, Le Chat, Meta AI, iAsk and Perplexity. Matomo says the feature is available in On-Premise from version 5.5.0.
Human referrals in the AI Assistants report
Open Acquisition > AI Assistants to see total visits, row evolution, goal conversions and segmented visit logs. Use landing pages and goals to answer whether a particular assistant sends qualified visitors, not merely whether it sends volume. Segmented logs are useful for checking the actual referrer and session behavior behind an aggregate row.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesSeparating crawlers and agents
Matomo provides separate AI Agent and AI Chatbot reports for automated systems that retrieve or request content. This distinction prevents a crawler’s page requests from being interpreted as human engagement. Matomo explicitly treats AI Assistant attribution as visitors referred by assistants; direct AI-tool access belongs in the agent or chatbot reports.
Matomo’s historical limitation
The dedicated channel applies to data collected after the relevant product update. Older records may not be reclassified into the new channel, so a year-over-year chart can show a measurement break. Mark the deployment date in your reporting notes and avoid comparing pre-channel data with post-channel data as if the collection rules were identical.
Choose Matomo Cloud when you want hosted deployment. Choose On-Premise when data location, infrastructure control or internal policy requires self-hosting; confirm that your installation is running version 5.5.0 or newer for the AI Assistant channel.
Adobe Analytics: enterprise analysis of conversational AI referrers
Adobe Analytics is most suitable when your organization already uses Analysis Workspace, Adobe data governance and custom dimensions. In Workspace, the Referrer type item Conversational AI tools is backed by a maintained list that includes domains such as chatgpt.com.
Why Adobe’s model is useful
Adobe’s documentation distinguishes pretraining scraping, on-demand prompts that may retrieve raw HTML without JavaScript, and agentic browser workflows. Clicks on source links in AI answers and agentic browser workflows can generate analytics hits. Treat those as different populations when you investigate traffic quality: a raw-HTML fetch can be technically real but cannot produce a browser session, click path or conversion event in the usual way.
Practical Workspace workflow
- Build a segment using Referrer type equals Conversational AI tools.
- Break the segment down by referrer, landing page, device, geography and your conversion dimensions.
- Compare visits, engagement and revenue with the Referral channel over the same period.
- Use server or CDN logs to investigate requests that Adobe does not see because JavaScript was not executed.
Adobe’s advantage is configurability: enterprise teams can combine the conversational-AI referrer type with product, account, campaign and revenue dimensions already used in Workspace. Its trade-off is implementation and administration effort compared with a lightweight analytics product.
Plausible: simple AI-source reporting with privacy-focused implementation
Plausible automatically groups known assistants into an AI Assistants channel. Its documentation names ChatGPT, Claude, Perplexity, Gemini and Copilot among the recognized sources.
What you can measure
Use the channel to connect AI-source visits with landing pages, engagement, goals, funnels, signups, orders, revenue and average order value. This makes Plausible suitable for a site that wants a small, understandable report rather than a large attribution configuration.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Human-visit focus
Plausible states that known AI crawlers and bots are filtered. The AI Assistants channel is therefore intended to represent people who click through, not every automated request from an AI infrastructure provider. Unknown automation can still require log-level investigation, so treat filtering as a documented rule rather than proof that every remaining session is human.
When Plausible is a good choice
- You want automatic grouping without maintaining GA4 regular expressions.
- You need conversion and revenue context but not a large enterprise workspace.
- Privacy and a minimal implementation are more important than extensive cross-product modeling.
Fathom: AI-source reports with historical data and exports
Fathom’s Sources tab includes an AI bucket. It identifies tools such as ChatGPT, Gemini, Perplexity, Claude, Microsoft Copilot, DeepSeek, Grok, Meta AI, You.com and Poe.
Capabilities documented on October 2, 2026
Fathom’s October 2, 2026 changelog says the AI tab works on historical data, supports per-tool filtering, allows CSV export and exposes an API dimension named ai_source. Historical reclassification is important when you need to examine a trend that began before the feature was introduced.
How Fathom assigns source data
Fathom combines referrers with UTM-tagged links. A visit that has neither a referrer nor a UTM source appears as Direct. If you control links distributed in an assistant, consistent UTM tagging can preserve attribution when an app strips the referrer; it cannot identify an assistant when the link is copied without its parameters.
Useful Fathom workflow
- Open Sources and select AI.
- Filter to one assistant to compare landing pages, engagement and conversions with the full AI bucket.
- Export CSV data for a finance, content or growth report, or query the API using
ai_source. - Investigate unusually large Direct increases alongside referrer and server-log changes.
Comparison by the decisions that matter
Human referrals versus automated access
| Question | GA4 | Matomo | Adobe Analytics | Plausible | Fathom |
|---|---|---|---|---|---|
| Does the vendor document a separate machine view? | Not as a dedicated AI-agent report; it classifies observed referral traffic | Yes: AI Agent and AI Chatbot reports | Yes, with distinctions for scraping, raw-HTML retrieval and agentic workflows | Known AI crawlers and bots are filtered | Known bots are filtered; unresolved traffic may be Direct |
| Best validation source | Referrer details and server logs | AI Assistant plus AI Agent/Chatbot reports | Workspace segments plus server/CDN logs | Channel and goal reports | AI source reports, exports and logs |
Assistant coverage and maintenance
Automatic channels reduce setup but depend on each vendor’s recognition list and update schedule. Matomo names eight assistants; Plausible names five examples; Fathom names a broader set including DeepSeek, Grok, You.com and Poe. GA4’s built-in channel covers recognized referrers, while custom regex rules let you add patterns but create a maintenance obligation. Adobe’s maintained Conversational AI tools list is appropriate for an enterprise that wants a managed classification inside Workspace.
Conversions, revenue and data access
- GA4: Uses GA4 events, conversions, audiences and ecommerce reporting.
- Matomo: Provides goal conversions, visit logs and row evolution in the AI Assistants report.
- Adobe: Combines conversational-AI referrers with configurable Workspace dimensions and enterprise revenue models.
- Plausible: Connects AI visits to goals, funnels, signups, orders, revenue and average order value.
- Fathom: Adds historical data, per-tool filters, CSV export and the
ai_sourceAPI dimension.
Privacy and deployment
| Deployment question | GA4 | Matomo | Adobe Analytics | Plausible | Fathom |
|---|---|---|---|---|---|
| Typical operating model | Hosted Google analytics property | Cloud or self-hosted On-Premise | Enterprise hosted analytics ecosystem | Privacy-focused hosted analytics | Privacy-focused hosted analytics |
| Most important trade-off | Broad ecosystem, but more configuration and governance | Control and machine reports, with deployment responsibility for On-Premise | Powerful customization, with higher implementation complexity | Simple reporting, with less enterprise-level customization | Simple reporting and exports, with source loss when referrer and UTM data are absent |
A reliable implementation and validation plan
Use the following sequence regardless of the product you select. It keeps referral reporting separate from assumptions about what an AI model may have done.
- Define the outcome. Decide whether you need visits, engaged sessions, leads, purchases, revenue, content assists or machine-request volume. Do not combine human and automated objectives in one KPI.
- Record the recognition rules. Note the product version, assistant domains and the date the channel became available. This is essential for Matomo’s post-update limitation and for GA4 custom regex maintenance.
- Verify tagging. Confirm that the analytics tag runs on every relevant landing page and that conversion events fire once. For links you control, use consistent UTM parameters as a backup to referrer attribution.
- Create a baseline. Save a report containing AI visits, landing pages, engagement, conversions and revenue before changing channel rules. Include Direct and generic Referral for comparison.
- Test a known click. Open a published assistant answer, click the source link in a normal browser session and confirm the expected referrer, channel and landing page. Do not treat a manually typed or copied URL as an attribution test.
- Check machine requests separately. Compare analytics with web-server or CDN logs. A request that fetched HTML without executing JavaScript will appear in logs but not as a normal analytics session.
- Annotate changes. Record channel-rule edits, vendor updates, consent-banner changes, redirect changes and site migrations. A sudden jump can reflect measurement changes rather than demand.
- Review by assistant and outcome. Rank sources by qualified conversions or revenue, not just sessions. A small source with high-value customers may deserve more content attention than a high-volume source with no outcomes.
Why AI traffic appears as Direct
Direct does not necessarily mean a visitor typed your address. An assistant app may open a link in a webview that omits the referrer; a browser may apply a privacy policy; a person may copy the URL; or the link may lack UTM parameters. Fathom explicitly places visits with neither a referrer nor a UTM source in Direct. The same underlying attribution loss can affect other tools.
- Compare Direct spikes with assistant launches, referral changes and server logs.
- Preserve UTM parameters through redirects and canonical-link systems when you distribute tagged URLs.
- Do not reclassify all Direct traffic as AI traffic without corroborating evidence.
- Report a range or a qualified trend when referrer loss is substantial, rather than claiming an exact assistant share.
Choosing in common scenarios
You already run GA4 across the company
Start with the built-in AI Assistant channel, then add a custom channel group only for assistants or domains that are not recognized. Keep the custom channel above Referrals and review conversion events and ecommerce values in the same reports used for other acquisition channels.
You need to distinguish people from crawlers
Matomo is the most explicit choice because its AI Assistants, AI Agent and AI Chatbot reports are separate. Adobe is also appropriate when your enterprise needs to analyze scraping, raw-HTML retrieval and agentic workflows alongside configurable business dimensions.
You want a small privacy-focused stack
Plausible provides an automatic AI Assistants channel and connects it to goals, funnels and revenue. Fathom is a strong alternative when historical reporting, per-tool filters, CSV exports or the ai_source API field are priorities.
You need enterprise-grade customization
Adobe Analytics fits teams that already govern data in Analysis Workspace and need conversational-AI referrers combined with product, account and revenue dimensions. GA4 can cover many organizations, but Adobe’s configurable workspace is better suited to complex enterprise taxonomies.
Measurement limits you should state in reports
Every product in this comparison measures what reaches the site with enough information to classify it. None can see an assistant answer that mentions your brand but generates no click, and none can reliably identify a visit after a referrer has been removed. Automated retrieval may bypass JavaScript entirely. For important decisions, reconcile browser analytics with server or CDN logs and use tagged links where you control distribution.
Free tools Windows power users keep installed
One-click scans. No signup required.
Because vendors maintain different recognition lists and apply updates at different times, do not compare raw AI-assistant percentages across products as if they were standardized measurements. Use one tool consistently for trend reporting, document its rules and treat product changes as potential breaks in the series.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
