Visitors, visits and views: Umami’s five core metrics explained

Visitors tell you how many distinct visitor identities Umami counted. Visits tell you how many visits they made. Views tell you how many page views were recorded. One person can contribute to all three numbers more than once, so they answer different questions.

This guide describes the website overview in Umami 3.4.1, with the normal page-view tracker and no filters. Event reports, older versions and custom tracking setups can use different counting rules.

Five metrics, five questions

Metric	The question it answers	What it does not tell you
Visitors	How many distinct visitors were counted?	The exact number of people across all their devices
Visits	How many separate visits were recorded?	How many new people arrived
Views	How many page views were recorded?	How many different pages exist, or how many buttons were clicked
Bounce rate	What share of visits had one page view and no custom event?	Whether every one-page visit was unsuccessful
Visit duration	What was the average measured visit length?	How long people actively read or watched

Read the numbers for the same website, date range, time zone and filters. Otherwise, even correct numbers can tell a misleading story.

Visitors: an estimate of your audience

Umami counts distinct visitor identifiers within the selected period. For anonymous tracking, it derives these identifiers from information including the website, IP address and browser user agent, with a rotating salt. That does not require storing the raw IP address, but it also does not produce a permanent identity for a person.

Someone switching devices, browsers or networks can be counted again. People with matching identifying inputs can be grouped together. The default salt rotates monthly in this version, so do not treat a long-range visitor total as a lifetime count of unique people.

There is another easy trap: do not add daily visitor totals to calculate weekly visitors. The same visitor may appear on several days. Select the full week and let Umami count the distinct visitors for that period.

Visits: one visitor can come back

A visit groups recorded activity under a visit identifier. A visitor might browse in the morning and return in a separate visit that afternoon. You can therefore have more visits than visitors without gaining a single new reader.

Use visits to understand how often your site is being used. More visits per visitor can suggest repeat use, but this ratio alone does not prove retention or satisfaction. Visit boundaries depend on Umami’s version and tracking implementation; do not assume they match another analytics tool’s inactivity rule.

Views: count page views, not clicks

In this version’s normal website overview, Views counts page views. Reloading a page can add another view when the tracker sends another page-view request. A correctly tracked client-side route change can do the same.

Custom events are not extra page views in this overview. Recording a download-button click does not, by itself, increase Views. Performance measurements are excluded too. Use the event report to answer how often a tracked button was clicked.

Views can rise because more people arrived, because people explored more pages, or because pages were reloaded. Read Visitors and Visits alongside Views before choosing an explanation.

Bounce rate: one page is not always a failure

For the unfiltered overview covered here, a bounce is a visit with one page view and no custom event. Bounce rate is the number of bouncing visits divided by the number of visits, expressed as a percentage.

A reader who opens one article, finds the answer and leaves may still count as a bounce. Someone who opens the same article and triggers a tracked download click does not meet that definition, even without opening another page.

This means adding event tracking can lower bounce rate without changing how people behave. Before celebrating an improvement, check whether you changed the tracking code. Do not compare this metric directly with another tool’s engagement-based bounce rate.

Visit duration: a measured interval, not a stopwatch

The overview measures the time between the first and last recorded page view in each visit, adds those intervals, then divides by the number of visits.

A visit with only one page view has a measured interval of zero. Someone could read that page for ten minutes, but there is no second page view to mark the end. The time spent on the final page is not measured by this calculation either. A custom click can change the bounce classification without adding a second page-view timestamp.

Treat duration as a trend with measurement limits, not as exact reading time. A longer average can mean deeper exploration, but it can also mean someone struggled to find what they needed.

A worked example: two visitors, three visits

The following is an illustrative example, not Okayu traffic data. Assume both visitor identities remain stable that day, all requests arrive, no filters apply, and Umami records the afternoon return as a separate visit.

Recorded visit	Page views	Custom event	Measured duration
Visitor A at 09:00	Home at 09:00, article at 09:01, pricing at 09:02	None	120 seconds
Visitor B at 10:00	Article at 10:00	Download click at 10:00:30	0 seconds
Visitor A returns at 15:00	Article at 15:00	None	0 seconds

The overview would show:

• Visitors: 2. A returns, but is still the same visitor in this example.
• Visits: 3. A has two visits and B has one.
• Views: 5. Three plus one plus one page views. The click is not a sixth view.
• Bounce rate: about 33%. Only A’s afternoon visit has one page view and no custom event: 1 ÷ 3 × 100.
• Average visit duration: 40 seconds. The measured intervals total 120 seconds: 120 ÷ 3. This does not mean each person spent 40 seconds reading.

That one download click prevents B’s visit from being a bounce, but it still does not prove a download or installation happened. It proves only that the click was recorded.

A useful daily check

1. Choose one website, a date range and a time zone. Clear filters you no longer need.
2. Start with Visitors: did your measured audience change?
3. Read Visits and Views together: did people return or view more pages?
4. Compare bounce rate and duration with a comparable period. Check for tracking changes before interpreting a jump.
5. Narrow the view by page, source or device to investigate. Treat an unusual number as a question, not a conclusion.

Avoid comparing a partial day with a finished day and calling the difference growth. Compare equivalent periods and remember that ad blockers, failed requests and your own traffic exclusions can all affect what is recorded.

Read the same metrics on iPhone and iPad

Okayu displays these five metrics from your connected Umami server, with trends, date selection, time zones and period comparisons. It is useful for a quick check away from your computer; it does not install tracking code or replace Umami’s website administration.

Use the same website, date range, time zone and filters when comparing the app with the web dashboard. Older server versions can have different metric definitions. If something looks wrong, first check those settings and the time of the last refresh rather than assuming one screen is counting a different audience.