Why analytics tools disagree about visitors

Measure records pageviews and registered custom events. A pageview becomes one pageview when the tracker sends it. Page reloads and client-side navigation can create more pageviews. The tracker sends the page path without the query string or fragment, so a campaign parameter does not create a separate page path.

For a Measure pageview, the server creates a keyed hash from the project, source, remote IP address, user agent, and that day's secret salt. It stores the hash only as needed to prevent a second count during that UTC day. It does not store the raw IP address, raw user agent, raw hit, or a persistent visitor identifier.

The salt changes each day. Measure cannot join those daily hashes. It cannot report how many people returned next week, a lifetime visitor count, or a returning-user rate. Read a Measure visitor total as distinct daily hash inputs that sent accepted pageviews. Do not translate it to known people.

Measure starts a visit with a pageview and begins another visit after more than 30 minutes without activity. It keeps this visit state in memory. A second pageview inside that window adds a pageview but does not start another visit. A later pageview starts a new visit.

One daily visitor can have several visits. A person can also count as more than one visitor when the hash inputs differ. Conversely, several people can produce the same inputs. Read visits as grouped activity, pageviews as loads, and visitors as a daily estimate. Do not use one as a substitute for another.

Compare like with like. If another dashboard uses a 30-minute session, compare Measure visits with that dashboard's sessions. If it uses a persistent identity, do not expect its users or returning users to match Measure visitors.

Definitions are only the first difference. The Measure script sends a pageview from the visitor's browser to the Measure endpoint. A blocked script, a browser setting, a network failure, or a page that does not load the script leaves no Measure pageview. Another tool can collect a request that Measure did not see, or miss a request that Measure saw.

Time zones also change daily buckets. Measure's daily visitor salt and day rollups use UTC. A dashboard that labels days in a local time zone can place activity near midnight in a different day. Compare a full date range before you investigate one unusual day.

Scope matters too. Measure ties a source to one site. If another report combines a production domain, preview domain, and documentation domain, it can show a larger total. Compare the same hostnames, date range, environment, and traffic filters.

Measure reads the browser's document.referrer, reduces it to the referring host, and stores that host in its rollup. It records direct when the browser supplies no referrer or when the referrer is one of the site's own allowed domains. It does not keep the referrer path, query string, or fragment.

A direct total does not prove that every visitor typed the address. It means Measure did not receive an external referring host for those pageviews. A bookmark can produce an empty referrer. A site or browser can also apply a referrer policy that omits referrer information. See the MDN sources below.

Match the hostname, source, date range, and time zone. Compare pageviews first. Then compare sessions, identifiers, reset periods, and blocked or filtered collection. A difference can show which behavior the tools measure differently. Do not average the totals.