What Is a Session in Web Analytics?
How sessions are defined, the 30-minute timeout convention, what resets them, and how cookieless tools build sessions without cookies.
A session is one continuous visit: the pageviews and events a visitor generates from arrival until they stop. 'Stop' is the engineered part — nothing in HTTP announces that someone left, so analytics infers it with a timeout, conventionally 30 minutes of inactivity.
The standard mechanics
- A visitor's first hit opens a session; each subsequent hit extends it.
- 30+ minutes of silence closes it; the next hit starts a new session.
- Some tools also reset at midnight or on campaign change (a returning click with new UTMs starts a fresh session, preserving campaign separation).
Sessions without cookies
Classic tools stored a session ID in a cookie. Cookieless analytics derives session continuity differently — Clycyo uses sessionStorage (which lives only until the tab closes and never follows the user across sites) or time-windowed grouping server-side. The practical outcome is equivalent: hits cluster into visits, with no persistent identifier left behind. The deeper mechanics live in the cookieless explainer.
Session metrics and their traps
- Session duration is measured first-hit-to-last-hit, so a single-page visit of any length records as ~0 — making average duration systematically wrong on content sites. Active-time measurement fixes this.
- Sessions vs visitors: one person, three visits today = one visitor, three sessions. Conversion rates need a declared denominator — per session and per visitor answer different questions.
- Tab-restore artifacts: browsers resurrecting twenty tabs on Monday morning create phantom sessions; good trackers wait for actual visibility.
Sessions remain the right unit for behavior-within-a-visit questions; for anything spanning visits — attribution, retention, revenue — the visitor record is the unit that matters.