Traffic channels and sources

How Privatus Analytics classifies every visit into a traffic channel from UTM parameters and the referrer, including AI assistants and custom channel rules.

# Traffic channels and sources

> How Privatus Analytics classifies every visit into a traffic channel from UTM parameters and the referrer, including AI assistants and custom channel rules.

Every visit gets one **channel**, from the landing page's campaign
parameters and the referrer. Rules are checked in this order, and the first
match wins.

| Channel | Rule |
|---|---|
| `paid_search` | A paid click id (gclid, msclkid…) or paid `utm_medium` from a search engine |
| `paid_social` | A paid click id or paid `utm_medium` from a social network |
| `paid_other` | Any other paid traffic. Paid mediums: cpc, ppc, paid, paidsearch, paid_search, cpm, cpv, display, banner, affiliate, paid_social, paidsocial |
| `email` | `utm_medium` containing email, or `newsletter`. Webmail: mail.google.com, outlook.live.com, outlook.office.com, mail.yahoo.com, mail.proton.me, app.fastmail.com, mail.aol.com |
| `ai_assistants` | Any `.ai` referrer, chatgpt.com, chat.openai.com, perplexity.ai, claude.ai, gemini.google.com, bard.google.com, copilot.microsoft.com, you.com, phind.com, meta.ai, grok.com, chat.mistral.ai, chat.deepseek.com, poe.com. Also utm_source chatgpt, chatgpt.com, openai, perplexity, claude, gemini, copilot, grok, deepseek, mistral |
| `organic_search` | google, bing, duckduckgo, yahoo, yandex, baidu, ecosia, qwant, startpage, brave, naver, seznam, ask, aol, search.yahoo, kagi, mojeek, sogou, so.com. Also utm_medium organic, search |
| `organic_social` | facebook.com, instagram.com, t.co, twitter.com, x.com, linkedin.com, lnkd.in, reddit.com, youtube.com, tiktok.com, pinterest.com, snapchat.com, threads.net, bsky.app, mastodon.social, news.ycombinator.com, whatsapp.com, telegram.org, discord.com, quora.com, tumblr.com, vk.com, weibo.com, producthunt.com. Also utm_medium social, social_media, sm, social-network |
| `internal` | A referrer on one of the site’s own domains |
| `referral` | Any other referrer or `utm_source` |
| `direct` | No referrer and no campaign parameters |

Click ids (`gclid`, `gbraid`, `wbraid`, `msclkid`, `fbclid`, `ttclid`,
`li_fat_id`, `twclid`, `dclid`) mark a visit as a paid click. They're kept
only as a yes/no flag. The id itself is discarded.

## Referrers, sources and campaigns

- **Referrer** is the hostname the visitor came from, with `www.` and `m.`
  removed. The referrer's query string is never stored. Referrers from the
  site's own domains are *internal* and don't start a new source.
- **UTM parameters** (`utm_source`, `utm_medium`, `utm_campaign`,
  `utm_content`, `utm_term`) come from the landing page URL.
- **`ref`** is taken from `ref`, `via` or `source` query parameters.

A visit keeps the source it started with (first touch): later pages of the
same visit don't change it.

## Custom channel rules

**Site settings → Channels** lets you add your own rules, checked before
the built-in ones. Each rule has a channel name and one or more conditions
(all must match) on:

- `referrer_host`, `utm_source`, `utm_medium`, `utm_campaign`, `ref`, `path`
- with `is`, `contains`, `starts_with` or `regex` (case-insensitive).

Example: channel **Partners** when `ref starts_with partner-`. Rules apply
to new visits. The preview shows how recent traffic would be classified.

以 Markdown 檢視

Every visit gets one channel, from the landing page's campaign parameters and the referrer. Rules are checked in this order, and the first match wins.

Channel	Rule
`paid_search`	A paid click id (gclid, msclkid…) or paid `utm_medium` from a search engine
`paid_social`	A paid click id or paid `utm_medium` from a social network
`paid_other`	Any other paid traffic. Paid mediums: cpc, ppc, paid, paidsearch, paid_search, cpm, cpv, display, banner, affiliate, paid_social, paidsocial
`email`	`utm_medium` containing email, or `newsletter`. Webmail: mail.google.com, outlook.live.com, outlook.office.com, mail.yahoo.com, mail.proton.me, app.fastmail.com, mail.aol.com
`ai_assistants`	Any `.ai` referrer, chatgpt.com, chat.openai.com, perplexity.ai, claude.ai, gemini.google.com, bard.google.com, copilot.microsoft.com, you.com, phind.com, meta.ai, grok.com, chat.mistral.ai, chat.deepseek.com, poe.com. Also utm_source chatgpt, chatgpt.com, openai, perplexity, claude, gemini, copilot, grok, deepseek, mistral
`organic_search`	google, bing, duckduckgo, yahoo, yandex, baidu, ecosia, qwant, startpage, brave, naver, seznam, ask, aol, search.yahoo, kagi, mojeek, sogou, so.com. Also utm_medium organic, search
`organic_social`	facebook.com, instagram.com, t.co, twitter.com, x.com, linkedin.com, lnkd.in, reddit.com, youtube.com, tiktok.com, pinterest.com, snapchat.com, threads.net, bsky.app, mastodon.social, news.ycombinator.com, whatsapp.com, telegram.org, discord.com, quora.com, tumblr.com, vk.com, weibo.com, producthunt.com. Also utm_medium social, social_media, sm, social-network
`internal`	A referrer on one of the site’s own domains
`referral`	Any other referrer or `utm_source`
`direct`	No referrer and no campaign parameters

Click ids (gclid, gbraid, wbraid, msclkid, fbclid, ttclid, li_fat_id, twclid, dclid) mark a visit as a paid click. They're kept only as a yes/no flag. The id itself is discarded.

Referrers, sources and campaigns#

Referrer is the hostname the visitor came from, with www. and m. removed. The referrer's query string is never stored. Referrers from the site's own domains are internal and don't start a new source.
UTM parameters (utm_source, utm_medium, utm_campaign, utm_content, utm_term) come from the landing page URL.
ref is taken from ref, via or source query parameters.

A visit keeps the source it started with (first touch): later pages of the same visit don't change it.

Custom channel rules#

Site settings → Channels lets you add your own rules, checked before the built-in ones. Each rule has a channel name and one or more conditions (all must match) on:

referrer_host, utm_source, utm_medium, utm_campaign, ref, path
with is, contains, starts_with or regex (case-insensitive).

Example: channel Partners when ref starts_with partner-. Rules apply to new visits. The preview shows how recent traffic would be classified.