CQI Referrer Attribution

No. Classification is entirely server-side. No tracking scripts are added to the front end.

No. Session tokens are one-way cryptographic hashes (HMAC-SHA256) of an anonymized IP address and the current date. Raw IP addresses are never stored. Tokens cannot be reversed to identify an individual.

Not at the same time, if CQIP Site Services has its own Referrer Attribution module active. Both write to the same database tables, since this plugin was extracted from that module and preserves its table and option names for migration compatibility. If CQIP_SS_ENABLE_REFERRER is set to true in your wp-config.php , this plugin detects that and refuses to activate, showing an admin notice with the exact line to add to wp-config.php to resolve it. Existing attribution data carries over automatically once the conflict is resolved.

When a visitor arrives via a referrer that matches an entry in the AI Referrer Taxonomy, the session is classified as the AI Tools channel. The matched platform name (e.g. “ChatGPT”) is stored as the source. The default taxonomy includes 19 domain entries covering 16 distinct platforms, since ChatGPT, Google Gemini, and Microsoft Copilot are each matched against two domains (for example, both chatgpt.com and openai.com resolve to ChatGPT). The full default set: ChatGPT, Google Gemini, Perplexity, Claude, Microsoft Copilot, You.com, Phind, Poe, Character.AI, Hugging Face, Mistral, Groq, Grok, Meta AI, NotebookLM, and DeepSeek. The taxonomy can be edited from the AI Taxonomy tab.

The plugin does not store personal data and is registered with the WordPress personal data export and erasure framework. When CQI Consent is active, attribution is only persisted after marketing consent. As with any tool that processes visitor data, your own legal assessment of your site’s practices is advisable.

Visits with no referrer and no UTM parameters are classified as the direct channel.

Yes. Go to Attribution > Settings > Excluded channels and check any channels you want removed. Excluded channels are removed from the Dashboard summary cards, the channel share chart, and the Sessions report, including CSV export. This is commonly used to exclude Direct / Unknown so reports focus on attributable traffic.

By default, nothing is removed. Your attribution data, settings, and AI taxonomy stay in the database in case you reinstall later. If you want a clean removal, enable Attribution > Settings > Data and Storage > Delete data on uninstall before deleting the plugin. With that enabled, deleting the plugin permanently removes the attribution log, all plugin settings, and the AI taxonomy. This cannot be undone.

Go to Attribution > Settings > Path Exclusions . Add one path prefix per line. Any request whose URL starts with a listed path is silently dropped before classification runs. Common entries are /robots.txt , /sitemap.xml , /feed/ , and any path your monitoring tools or uptime checkers hit. The plugin also has a built-in guard that automatically excludes /xmlrpc.php , /wp-cron.php , /wp-signup.php , /favicon.ico , and other WordPress system paths regardless of your exclusion list.

When more than one signal is present on a session, this priority order applies:

1. 01UTM parameters (utm_medium determines email, social, paid search, paid display, or campaign)
2. 02AI Referrer Taxonomy match
3. 03Known search engine referrer (organic search)
4. 04Known social platform referrer
5. 05Referral (any other external domain with a referrer)
6. 06Direct (no referrer, no UTM parameters)

Both are detected from the utm_medium URL parameter, not from the referrer header. utm_medium set to cpc or ppc is classified as Paid Search. utm_medium set to display is classified as Paid Display. Any other utm_medium value is classified as Campaign.

If your paid traffic does not carry UTM parameters, those sessions are classified by referrer instead, typically landing in Organic Search, Referral, or Direct depending on the source. Tagging every paid URL with UTM parameters is required for accurate Paid Search and Paid Display classification.