Entry pages is the landing-page decision report, not a bare count. Each row carries the share of its landing visits that bounced, their average visible engaged seconds, and, when your orders carry the visit's landing path, the net revenue and order count that started there, per currency. One row answers "is this landing page worth the spend" by itself.

Exit pages rank by exit rate, exits from a path divided by every view of that path, because a raw exit count mostly ranks your most popular pages: everybody leaves eventually, and the interesting page is the one people leave disproportionately.

Traffic patterns: hours and weekdays

The Traffic patterns screen folds a range onto the calendar: the daily movement, the shape of a week, and the audience's day clock. Window widths are not here. They are a breakdown on Tech.

Hours folds any range onto one clock: 24 rows, one per local hour, whether you asked about a day or a year. Empty hours are returned as zero, because a count of zero at 4 a.m. is a measurement, not a gap. An hour can sum above the day's own total: a visit spanning two hours counts in both, the same way a visit spanning the site's local midnight can count in both days. New visits do not span midnight UTC.

Select a weekday bar on Overview or Traffic patterns to narrow the panel to that day of the week. The global Filter menu also offers Local weekday and Local hour. Both use your site's timezone and combine with other filters, so Monday at 09:00 stays Monday at 09:00 wherever you open the panel. Orders and refunds follow their own transaction dates. Imported daily totals and Search Console cannot answer this site-local segment and explain the limitation.

Journeys and funnels

Journeys discovers the paths visits actually took, as a flow you can click through: pick a node and see what followed it. It needs no setup beyond switching its module on. The screen reads a thousand patterns at a time, and its Listed patterns tile prints the real count below that, and "1,000+" at it, rather than a cap dressed as a measurement.

Funnels is the opposite direction: you declare the sequence and the report answers how far people got. Building one, on the Funnels screen:

Two to twelve steps, each an exact page path like /checkout or an exact event name like signup, and the two kinds mix freely in one funnel. Type the value in. Nothing preloads a list of every URL your site ever had. A path step is saved the way the collector spells a path, trailing slash off and percent-encoding decoded, and shown that way afterwards, so /checkout/ becomes /checkout and matches.
Counted per visit, in order: a visit enters at step one and is counted through every step it reached in sequence within that visit. Wandering between steps does not eject it. A later step counts only after its predecessor.
Owners and admins create, rename, and edit saved funnels. Viewers run them. What is saved is the definition alone, never a result, so a saved funnel always answers with current data over whatever range and filters you pick.

Clicking a page in either report is visit-scoped: it selects the visits that saw the page and keeps their whole sequences, so the answer shows real paths rather than the page confirming itself.

Core Web Vitals

With the performance module on, the Web Vitals screen joins the Behavior group and reports LCP, INP, and CLS from real visits, per page and per day, at the 75th percentile, because that is how the thresholds are defined. A lab tool answers what one machine saw once. This answers what the slower quarter of your actual readers experienced.

Vital	Measures	Good
LCP	Time until the largest content element renders	2.5 s or less
INP	Delay from an interaction to the next paint	200 ms or less
CLS	Unexpected layout movement, unitless	0.1 or less

Each percentile travels with its number of recorded measurement groups and their ratio to pageviews. This shows how much evidence is behind the result, with the midnight UTC limit explained below. Field data, never a lab score.

The browser's Web Vitals describe the document it loaded. A later route change in a single-page application does not produce a new largest paint. Kehai keeps those readings on the original view. Returning to the same URL or restoring a cached page cannot reuse its old measurements.

For current measurements, repeated readings from the same view form one observation, including after same-day inactivity. They need a matching pageview inside the selected range and filters. Its local date determines the daily bucket. The view's temporary code stays only in page memory, and its stored hash is scoped to the site and UTC day. An old view does not carry readings across midnight UTC. Older rows retain their previous visit-and-path grouping, whose coverage can exceed 100%.