Sources
The Referrers card — also called Sources — shows the top referrer hosts that sent traffic to your site in the selected range. Each row is the domain name extracted from the standard HTTP Referer header sent by the visitor's browser.
What a referrer host is
When someone clicks a link on another website to arrive at yours, the browser includes an HTTP Referer header in the request. That header contains the URL of the page the visitor was on before clicking. StatelessID extracts only the host portion — the domain name — discards the path and query string, and stores the host as the referrer class.
The host is normalised to lowercase and stored without any port number or path. A visit from https://news.ycombinator.com/item?id=12345 is stored as news.ycombinator.com. A visit from https://twitter.com/user/status/123 is stored as twitter.com. A visit from https://www.google.com/search?q=widgets is stored as www.google.com.
This normalisation means you see referrer hosts at the domain level, not the page level. If ten different Hacker News threads all link to your site, they all appear as news.ycombinator.com in one row. You cannot see which specific article sent the traffic without looking at a UTM parameter — but UTM parameters are stripped at collection time and not stored.
Direct traffic and the (direct) label
When no Referer header is present in the request, the visit is recorded with an empty referrer host. On the board, an empty referrer host is displayed as (direct). Direct traffic includes visitors who typed your URL into the address bar, clicked a bookmark, opened a link from an email client that strips the header, or followed a link in a native mobile app.
Direct traffic also includes visits where the Referer header was present in the browser but was not sent to your server. This can happen when the linking page uses a rel="noreferrer" attribute on the link, or when HTTPS → HTTP transitions strip the header by browser policy. There is no way to distinguish "truly typed the URL" from "header was stripped" using server-side data alone.
(direct) is often the largest single entry in the Referrers card, especially for sites with a large returning audience or a lot of email and app traffic. This is normal. It does not mean your analytics are broken.
Reading the ranked list
The Referrers card shows the top referrer hosts ranked by page view count for the selected range, with a horizontal bar proportional to share. The compact view shows the top four rows. The collapsible drawer does not offer an extended referrers list the way it does for page groups, but the CSV export includes the full referrer breakdown.
Rows below the privacy floor of five views are folded into Other. If your site receives traffic from many small referrers — niche blogs, small communities, individual social media posts — the Other row may be quite large. The named rows represent the largest individual sources.
The ranking is by visit count, not by engagement, revenue, or any other signal. A referrer that sends ten visits ranks above a referrer that sends two, regardless of what those visitors did on your site. StatelessID does not track what happened after the first page load, so engagement metrics are not available.
Campaign traffic and UTM parameters
StatelessID strips query strings from URLs before storing the path class. This means UTM parameters — utm_source, utm_medium, utm_campaign — are not stored and cannot be reported in a UTM breakdown. Campaign traffic is attributed to the referrer host only. A visit from a paid search ad on Google is attributed to the same google.com referrer host as an organic search click.
If you need campaign-level attribution, you would need a tool that stores UTM parameters. StatelessID does not store them because they appear in the query string, which is discarded at collection time. This is a deliberate design choice that keeps the path class clean and prevents query strings from fragmenting your page group data.
Unexpected referrer hosts
Occasionally you will see a referrer host in the list that you do not recognise. This can happen for several reasons. A referrer spam campaign sends beacons with fake Referer headers. A testing tool or monitoring service that loads your pages will have its own host as the referrer. A staging environment might have its own domain that sometimes appears in the list.
If you see a referrer host that sends a large number of visits but does not match any known source, check whether the visit count corresponds to a real traffic pattern in the by-day chart. A sudden spike from an unknown host with no corresponding visible traffic in the chart is a sign of beacon manipulation rather than real visits. The bot filter catches User-Agent-based bots but cannot catch all forms of referrer injection.
Troubleshooting
If (direct) is unexpectedly large compared to previous ranges, check whether you recently sent an email campaign or posted on a platform that strips referrer headers. This is a common explanation for a (direct) spike.
If a referrer you expect to see is missing, confirm it had at least five visits in the selected range. Also confirm that the linking pages do not use rel="noreferrer" on their links, which would strip the header before it reaches your server.