Google clarifies Search traffic identification in analytics and website logs

Google added a documentation page on November 12 about identifying search traffic in analytics systems and server logs. The update explains the use of the HTTP Referer header and the differences between visits recorded by a website and clicks reported by Search Console. The subject directly concerns the measurement of organic visibility and traffic reaching business websites.

According to Google, Search transmits the origin of the referring page, without its full path or address parameters. On the web and in the iOS application, this is Google's main domain or a local variant. The Android application can also transmit an application-specific value when it opens a result in a browser tab.

What information actually reaches the website

Referer is a header in the request sent to the server. It describes the address from which a page or resource was requested. The server can use it in logs, while analytics systems can interpret it to identify the source of a visit. The header's name retains the historical spelling Referer; the associated policy uses the differently spelled name Referrer-Policy.

The origin policy limits the information to the address's origin rather than the entire URL. Information about the referring site is therefore separated from the page path and parameter string. The no-referrer policy removes the header entirely. These behaviours are defined at the level of web technology and explain why a log can contain a shortened address or an empty field.

Google specifies that Discover can use the same values as the surface where it appears. Other services delivered from the same origins, such as Google Images, can also have the same signal. Identifying the referring domain does not automatically separate every Google experience into a distinct report.

A loaded page in turn generates requests for images, stylesheets and JavaScript files. For resources requested by that page, the header can indicate the website page using them. Requests for the landing page and requests for its components therefore represent different operations. The total number of HTTP requests in a log is not equivalent to the number of visits.

Clicks and sessions are measured at different moments

In Search Console, a click is recorded when someone selects a result leading to the website. The measurement of behaviour on the page happens later, after the destination loads and its collection mechanism executes. Google thereby distinguishes activity in search results from interactions on the site, two stages observed through different systems.

In Google Analytics, a session begins when a page or screen is viewed and no session is already active. The default setting ends a session after 30 minutes of inactivity, and that timeout can be configured. A session groups the user's activity; it is not a count of every click occurring before a page loads.

The session_start event generates the session identifier and session number. Attribution information can include the referrer, UTM parameters and the gclid identifier. Consequently, the same interaction is described through several types of data: the request observed by the server, the session in Analytics and the click observed in the search engine. Each indicator retains its own collection definition.

Traffic classification also uses campaign data

The Traffic acquisition report describes session sources for both new and returning users. The User acquisition report concerns the initial acquisition of new users. Dimensions with the Session prefix, such as session source and medium, answer a different question from dimensions tracking a person's first arrival on the website.

In Analytics documentation, source identifies where traffic originates, while medium describes the method of arrival, such as organic or cpc. UTM parameters place campaign data directly in the destination address. Auto-tagging through the Google Ads integration uses identifiers such as gclid and can supply details about the account, ad group, campaign and advertising network.

The default channel grouping has its own rules. Organic Search includes visits from non-advertising results, including AI Overviews and AI Mode. The AI Assistant channel covers sources such as ChatGPT, Gemini, Deepseek, Copilot or Grok and excludes those two Google Search experiences. In this classification, the name of an AI technology and the name of a traffic channel are not synonymous.

Search Console reports organic search clicks, whereas Analytics can also include paid traffic in its reports. A visit associated with Google therefore needs to be read alongside its medium and campaign information. A total combining organic and advertising sources describes a different population from the total organic clicks in Search Console.

Reports retain differences in addresses and calendars

Google also documents differences involving URLs: Search Console assigns data to the canonical address selected by Google, whereas Analytics can report any address where collection code is present. Two reports about the same website can therefore group activity under different addresses. For daily comparisons, Search Console uses Pacific Time, while the Analytics time zone is chosen in the property's configuration.

The Search Console performance report separates clicks, impressions, click-through rate and average position. Data can be grouped by query, page, country, device and date. Search queries are available through this reporting system; the header received by the website does not include the text a person entered into Search. A result's visibility and the visit recorded on the site remain different indicators.

The Landing page report in Analytics identifies the first page viewed in a session and can be combined with session source and medium. It includes sessions, users, engagement time and key events. These data connect the entry page with activity on the website, while information about the result's appearance and selection remains in search reporting.