---
title: "How to track traffic from ChatGPT and Perplexity"
url: https://hostmy.blog/docs/rankready/track-chatgpt-perplexity-traffic/
date: 2026-08-24
modified: 2026-08-25
lang: en
author: "Aditya Sharma"
description: "See visits that arrive from AI products, tell them apart from crawler hits, and understand what the numbers leave out."
word_count: 834
---

# How to track traffic from ChatGPT and Perplexity

RankReady tracks AI referrals: visits that arrive on your site from an AI product rather than from a search results page or a link on another site. This page covers where to find them, how they differ from crawler activity, and the gaps you should state whenever you report the numbers. No [AI provider](https://hostmy.blog/docs/rankready/choose-ai-provider-cost/) key is required.

## Referrals and crawler hits are different things

RankReady records both, and they answer different questions. Keep them apart in every report.

| | Crawler hit | AI referral |
| --- | ----------- | ----------- |
| Who made the request | An automated agent such as GPTBot or PerplexityBot | A person using a browser |
| What it proves | Your page was fetched | Someone clicked through to your site |
| Reaches your analytics | Usually not | Yes, like any other visit |

A referral is the stronger of the two, because a human took an action. Neither is a citation count.

## Where to look

Open the **RankReady** menu in wp-admin and go to the **Insights** tab. AI referral data and crawler activity are both surfaced there.

![The RankReady Insights tab in wp-admin showing AI referral and crawler activity](https://hostmy.blog/wp-content/uploads/2026/08/rr-insights-v2-scaled.png)Insights is where referral and crawler data are reported.

## Setup

There is nothing to connect. AI referral tracking is one of the features that works with no API key, alongside `llms.txt`, the Markdown endpoints, crawler controls, [Article schema](https://hostmy.blog/docs/rankready/add-article-speakable-schema/), the [author box](https://hostmy.blog/docs/rankready/how-to-add-eeat-author-box/), the freshness scanner and the [WebMCP manifest](https://hostmy.blog/docs/rankready/how-to-add-webmcp-wordpress/). Install, activate, and the data accumulates from that point forward. It is not retrospective, so the clock starts when you install rather than when you first look.

## How referral attribution works, and where it fails

A referred visit is identified from the referrer the browser sends when a person follows a link. When that information is present, the visit can be attributed to the product it came from. When it is absent, the visit cannot be attributed, and it will appear as ordinary direct traffic instead.

The referrer is missing more often than people assume. It can be stripped by the referrer policy of the page containing the link, by privacy settings and extensions, and by the route a link takes. Copy a link out of an answer and paste it into a new tab, and there is no referrer at all.

The consequence is straightforward and worth writing next to the number: **AI referral counts undercount, by an amount you cannot measure.** Treat them as a floor, never as a total.

## The bigger gap: answers without clicks

Referral tracking counts clicks. An assistant that summarises your page and names your site without the reader following a link produces no referral. Someone who reads the answer, remembers your brand and searches for you the next day arrives as a search visit or a direct one.

So a low referral count does not mean your content is unused, and a high one does not measure how often you were mentioned. It measures one specific, observable thing: people who clicked.

## What to report, and how to word it

Language keeps this honest. Two lines to use, and two to avoid:

- **Good:** "This month, named AI crawlers fetched these pages, and this many visits arrived with an AI product as the referrer."
- **Good:** "Referral counts are a floor, since attribution depends on a referrer being sent."
- **Avoid:** "We were cited N times." Your server cannot see citations.
- **Avoid:** percentages, multipliers and growth rates presented as industry facts. If a number did not come out of your own install, do not put it in the report.

## Cross-checking with your analytics

Your analytics platform can usually segment traffic by referrer too, and comparing the two is worth doing. Where the counts differ, the cause is normally mechanical rather than mysterious:

- Your analytics script is blocked for some visitors while the server still handles the request.
- Caching layers and CDNs change which requests reach WordPress at all.
- The two systems attribute and window sessions differently.

Use them as two imperfect views of the same traffic. Pick one as your reporting source, note the other as a sanity check, and do not add them together.

## If you see no referrals

- **Check how long tracking has been running.** A small site can go a long time between AI referrals. A zero on a short window is not a finding.
- **Check your caching.** If pages are served entirely from an edge cache, those requests may never reach WordPress, and a WordPress plugin cannot record what it never sees.
- **Check that your content is reachable.** Load `/llms.txt` and a `.md` URL yourself. If they 404, re-save permalinks under **Settings, Permalinks**.
- **Check your crawler rules.** Open `/robots.txt` and read the block between `# BEGIN RankReady` and `# END RankReady`. If you have disallowed the agents whose products you were hoping to see referrals from, the rules are working as configured.

## What this data is for

Referral tracking tells you whether people are reaching your site from AI products, and which pages they land on. That is a useful, bounded fact. It does not tell you how often you were mentioned, and no plugin can. Report the floor, name the gap, and let the number be what it is.