One reader, three pages
Imagine one reader opens your home page, follows a link to an article, then refreshes that article. If all three pageviews arrive under the same visitor and visit identifiers, that is one visitor, one visit and three views.
Now a second reader opens the same article using a different browser and network. Assuming a distinct identifier is created, the totals become two visitors, two visits and four views. These numbers tell different parts of the story: audience size, grouped activity and content consumption.
Reader A: home → article → refresh = 3 views
Reader B: article = 1 view
Total: 2 estimated visitors, 2 visits, 4 viewsWhat makes a visitor “unique”?
In our standard cookieless setup, the identifier combines the website, IP address and browser user-agent with a salt that rotates each month. Raw IP addresses are not saved in the analytics database. A stored tracking cookie is not needed to make this estimate.
This is not a verified count of human beings. Two people using the same browser type on a shared network can look alike. One person changing networks or browsers can look different. The monthly rotation also breaks continuity, so the same person can contribute more than one identifier across a month boundary.
Why page rows do not add up to your visitor total
Someone who reads your home page and your article appears among the visitors to both pages. Adding the two page rows counts that person twice. The overall Visitors figure counts distinct identifiers across the selected dataset instead.
The same issue applies to daily or hourly visitor bars. A reader present on two days may count once in each day and once across the combined period, when the identifier stays the same. Use the report’s total for the full range rather than summing unique counts from smaller buckets.
A visit depends on the measurement rules
Our Umami-based visit identifier uses an hourly salt and an in-memory cache with a 30-minute expiry. It is not the same as a conventional session cookie that resets after 30 minutes of inactivity. Changing the time range or restarting collection infrastructure can also affect how grouped activity appears.
When comparing tools, compare pageviews first: they are the most direct record of what each tag received. Then compare visitor and visit definitions, timezones, filters and date boundaries. A differently defined metric is not automatically a broken metric.