Traffic Sources

How a Source Is Named

Klar stores the visit's referrer and, when the pageview is recorded, one source name for it. The name is the referrer's host, with a leading www. removed and well-known hosts mapped to a single label: every Google domain is Google, news.ycombinator.com is Hacker News, and t.co is X. The path and query of the referrer are never part of the name.

Direct and Unknown Sources

• (direct): the visit had no referrer. That covers typed addresses, bookmarks, many apps and any browser that withholds the referrer. It is a source of its own, never merged into another.
• Unlisted hosts: a referrer from a site Klar has no label for appears as its own hostname (for example blog.example.org) and counts as a referral. Two spellings of the same host are one source.
• Not a host: a referrer value that is not a hostname is kept as it was received, so it stays visible instead of being merged into (direct).
• Sign-in redirects through accounts.google.com are not a referral and are recorded as (direct).

Drilling Into One Source

In the Sources panel, open the ⋯ menu on a row and choose Only this. The page filters to that source: the visitor and pageview totals, the chart, and every other panel (pages, campaigns, countries, devices) then show that source's traffic only. Exclude this does the opposite. The filter is the source parameter in the address bar, for example ?source=Hacker%20News, so a drill-down can be shared as a link.

The Filter button offers the same field with is, is not, contains and does not contain. Contains matches the source name shown in the panel. Actions follow the filter through their session; Goals, which carry a path only, report as unavailable.

Older Data and Imports

Pageviews recorded before the source name was stored are labelled by a one-off backfill run by the operator; until then they match no source filter (but are kept by is not). Imported history is matched on its own source values and is included when the source is the only filter.