Pageviews vs visits vs unique visitors: what each one counts

A pageview is a page that loaded, a visit is a run of them, and a unique visitor is whatever your tool uses to recognise a person. They drift apart for reasons you can read, once you know each tool's rules and why daily visitors never add up to a week.

11 min read

Summarize with

In this article
  1. 1Three numbers, three different things
  2. 2Pageviews: every page that loaded and reported
  3. 3Visits: where one ends is a rule, not a fact
  4. 4Unique visitors: a stand-in for a person
  5. 5Two readers, two days
  6. 6Why daily visitors don't add up to a week
  7. 7What GA4, Plausible, Matomo and Cloudflare say
  8. 8The ratios say more than the totals
  9. 9When they move apart
  10. 10Which number answers which question
  11. 11The short version

Most analytics dashboards lead with the same three numbers: pageviews, visits and unique visitors. They tend to rise and fall together, so it's easy to read them as three sizes of one thing. They aren't. Each one counts something different, and part of what it counts is a rule your tool chose.

Here is what each one counts, a small example worked through by hand, how GA4, Plausible, Matomo and Cloudflare Web Analytics define them, and what it means when the three move apart. The details about those tools come from their own documentation, as of November 2026.

Three numbers, three different things

In one sentence each:

  • A pageview is one page that loaded and ran the tracking script, which reported it.
  • A visit, also called a session, is one visitor's pageviews close together in time. Where one visit ends and the next begins is a rule the tool sets.
  • A unique visitor is whatever the tool recognises as the same visitor (a cookie, a token in browser storage or an anonymous key), counted once in the period you picked.

One visitor can make several visits, and one visit can hold several pageviews, so pageviews are normally the largest of the three and unique visitors the smallest. What's worth reading is how far apart they sit, and when that changes.

Pageviews: every page that loaded and reported

A pageview needs no judgement: a page loaded, the script ran, it counts. That makes it the steadiest of the three, and also the easiest to throw off without anyone noticing:

  • Reloads. A reload loads the page again, which is usually another pageview. Plausible, for example, defines its total as the number of times pages were loaded.
  • Single-page apps. A site that swaps pages without a full load needs a script that notices the address change. If it doesn't, every visit looks like one pageview. If two things report the same change, every page counts twice. The Next.js guide shows how to test it, and the test works on any framework: click through a few pages and watch the count.
  • The tag on a page twice. A builder's header field plus a theme's own copy, for instance. In a tool that doesn't guard against it, each copy reports the page, which can double pageviews while visitors stay where they were.
  • Pages that were seen but never reported. An ad blocker, a script that failed to load, or a declined cookie banner in a tool that waits for one.
  • Things that aren't people. Many crawlers don't run JavaScript, so a script-based tool never sees them. The ones that do are left to each tool's bot filter.
  • You. Your own visits count, unless your tool lets you leave them out.

Some tools add a fourth figure, unique pageviews, which counts a page once per visit. In Matomo's own example, two visitors who each open the home page four times make eight pageviews and two unique pageviews. It tells a page many people saw apart from a page a few people kept reopening.

Visits: where one ends is a rule, not a fact

A visit groups the pageviews that belong together. Someone who lands on your home page, reads two articles and closes the tab has made one visit with three pageviews.

The hard part is the end. Browsers don't reliably tell anyone when a tab closes, so tools use a rule instead, nearly always a stretch of inactivity. GA4, Plausible and Matomo all end a visit after 30 minutes without activity by default. Some add more rules, or use a different one altogether:

  • Matomo, according to its FAQ on visits, also starts a new visit when someone arrives through a new campaign, or keeps browsing past midnight in the site's time zone.
  • GA4 lets you change the timeout in the web data stream's tag settings, as Google's page on sessions describes.
  • Cloudflare Web Analytics defines a visit without a timeout: its documentation counts one whenever a page view comes from another website or a direct link, judged by whether the referrer matches the site's hostname, so its visits are closer to arrivals.

Every rule changes the count. A shorter timeout splits one long read into two visits, and a longer one merges two separate ones. That reaches past the visit total, because several figures belong to the visit rather than to the pageview: the page it started on, the source that sent it, and whether it was a bounce. Google says the event that starts a GA4 session carries what the session is attributed to, such as UTM parameters and the referrer. A visit with a single pageview is a bounce in the plain definition, though GA4 draws that line differently, as a bounce isn't a bad visit explains.

Unique visitors: a stand-in for a person

To count unique visitors, a tool has to decide that two pageviews came from the same person. It can't know that. It can only recognise something it has seen before, and the count is how many different somethings turned up in the period. There are three common kinds:

  • A cookie. Google says GA4's _ga cookie is there to tell users apart, and that it lasts two years by default. Clear it, or open a private window, and the same person is new.
  • A random token in browser storage. The same idea without a cookie, with the same weaknesses.
  • An anonymous key worked out from the request. Plausible's data policy describes a hash of a daily salt, the site's domain, the IP address and the user agent, with the salt rotated and deleted every 24 hours, so a key can't link one day to the next.

Each one miscounts in known ways. One person on a phone and a laptop is two visitors with any of them. A family sharing one computer can be one. With a daily key, the same person on two days is two. The four ways to count without cookies, and where each goes wrong, are in how to count unique visitors without cookies. What follows is about what that choice does to the numbers you read.

Two readers, two days

Take a made-up site with two readers over two days.

  • Reader A arrives on Monday at 09:00 and reads three pages in ten minutes. At 20:00 they come back and read one more. On Tuesday at lunch they read one page.
  • Reader B arrives on Tuesday evening and reads two pages.

Two readers, two days

Illustration, not real data

Monday

Tuesday

Reader A

09:00
20:00
12:30

Reader B

No visit
18:00
One pageviewOne visit
A made-up example. Every gap of more than 30 minutes starts a new visit, so reader A's Monday is two visits.

Count it with a 30-minute visit rule:

PeriodPageviewsVisitsUnique visitors
Monday421
Tuesday322
Both days, if the tool recognises A on Tuesday742
Both days, if the key changes every day743

Pageviews and visits add up: Monday's plus Tuesday's is the total for both days. Unique visitors don't. A tool that recognises reader A on Tuesday counts two people over the two days, fewer than the three you get by adding the days. A tool whose key changes every day can't link Monday's reader A to Tuesday's, so it counts three, exactly the sum.

And if reader A had used a laptop in the morning and a phone in the evening, Monday would show two visitors instead of one, in either kind of tool. Its pageviews and visits would stay the same, because the evening was a separate visit anyway.

Why daily visitors don't add up to a week

That's the rule worth remembering. Pageviews and visits can be added across days. Unique visitors can't, because someone who came on three days of a week is one visitor to a tool that recognises them, and appears in three of its daily counts.

Matomo's documentation warns about exactly this: adding up daily unique visitors doesn't give the total for the range. Each period's figure has to be worked out on its own, a query Matomo calls costly, so by default it does that only for set periods (days, weeks and months on its self-hosted version, days and weeks on Matomo Cloud) and not for years or custom date ranges.

With a daily key, the arithmetic flips. Plausible says a person who visits on several days, or from several devices, counts as separate visitors, while several visits on one day count once. In a tool like that, a week's unique visitors is close to the sum of its seven days, and someone who reads every day counts seven times.

Pageviews and visits add up across days. Unique visitors don't, unless the tool forgets everyone each day.

What GA4, Plausible, Matomo and Cloudflare say

The words differ from tool to tool, and so do the rules behind them. This is what each one's own documentation says, as of November 2026:

ToolPageviewsVisitsUnique visitors
GA4Views: the web pages and app screens users sawSessions: end after 30 minutes of inactivity by default, adjustable per web streamUsers, which in reports means active users, told apart by the _ga cookie
PlausibleTotal pageviews: every time a page was loadedTotal visits: end after 30 minutes without activityUnique visitors: separate days and separate devices count separately, several visits on one day count once
MatomoPageviews, plus unique pageviews that count a page once per visitVisits: a new one after 30 minutes of inactivity, a new campaign, or midnight in the site's time zoneUnique visitors: worked out per day, week and month by default (day and week on Matomo Cloud); adding up the days doesn't give the total
Cloudflare Web AnalyticsPage views: successful responses with HTML contentVisits: a page view whose referrer doesn't match the site's hostnameNot among the metrics it lists

The sources are GA4's list of metrics, Plausible's metric definitions, Matomo's FAQs linked above and Cloudflare's high-level metrics.

Put two of these on one site and all three numbers will differ, for the reasons in that table and several more: bot filters, ad blockers, consent banners and time zones. Google also says GA4 works out its session count by estimating the number of distinct session IDs, rather than counting each one. The rest is in why no two analytics tools agree on your traffic, and reports that are estimated rather than counted are in the guide to data sampling. The practical rule: compare a tool with itself over time, not with another tool.

The ratios say more than the totals

The three totals get more useful once you divide them.

  • Pageviews per visit (views per session in GA4, views per visit in Plausible) says how far a visit goes once it has arrived.
  • Visits per visitor (sessions per active user in GA4) says how often people came back within the period.

The second one only means what it says if the tool can recognise someone across the whole period. With a daily key, a second visit by the same visitor can only happen on the same day, so over a month the ratio tells you about same-day returns, not about loyalty. In the example above, the same four visits give 2.0 visits per visitor in a tool that recognises reader A on Tuesday, and about 1.3 in one whose key changes daily. The readers did exactly the same thing in both.

Read both ratios against their own history rather than another site's. A documentation site and a one-page landing page will look nothing alike, and both can be doing their job.

When they move apart

The gaps between the three numbers are where the information is. Four patterns, and what to check first:

Pageviews up, visits and visitors flat

Each visit is seeing more pages. People reading more is one explanation. A tag that fires twice, a page that reloads itself or a redirect in the browser that counts both pages are others. Look at which pages gained the views, and whether pageviews per visit jumped on one day.

Visitors up, pageviews per visit down

New people who read one page and leave: a link shared somewhere, a newsletter, an AI assistant's answer. Sources and entry pages show where they came from. If the source is Direct, what direct traffic really contains is the next read.

Visits up, visitors flat

The same people coming back more often within the period. Or the rule changed: a shorter session timeout splits visits that used to be one. If your tool lets you change the timeout, check that nobody did.

All three fall on the same day

Before looking for a reason in the market, check that the script is still on every page. A deploy can drop it, a new Content Security Policy can block it, and a new cookie banner changes what a consent-based tool sees. View the source of a page and look for the tag.

Which number answers which question

If you want to knowRead
How many people did this reach?Unique visitors, as an estimate, over a period your tool can recognise people across. With a daily key, read it day by day.
Which pages get read?Pageviews by page. Unique pageviews, where the tool has them, take out repeat views within one visit.
How far do visits go?Pageviews per visit, next to bounce rate.
Do people come back?Visits per visitor or returning visitors, and only with an identifier that lasts across days.
Where did people come from?Visits by source and by entry page, since the source belongs to the visit. Links you tag with UTM parameters show up there by name.
How much load is the server under?Server logs or your host's figures. Bots are load too, and a script-based tool misses every one that doesn't run JavaScript.

The short version

  • Pageviews are page loads, visits are runs of them, unique visitors are a stand-in.
  • Find your tool's rule for ending a visit and for recognising a visitor.
  • Don't add daily unique visitors into a weekly or monthly figure.
  • Read pageviews per visit and visits per visitor against their own history.
  • When the numbers move apart, check sources, entry pages and the script first.
  • Compare a tool with itself over time, not with another tool.

Lomen counts all three the plain way. A pageview is each page its script reports, page changes in single-page apps included, with known bots and pageviews from other domains left out. A visit is one visitor's pageviews split on a 30-minute gap, a gap that isn't a setting, and a new campaign doesn't start a new one. The standard script sets no cookies and stores nothing in the browser, Lomen never stores IP addresses, and visitors are counted from an anonymous key that changes daily. So a week's visitors is close to the sum of its days, and the dashboard calls that figure Visitors rather than Unique visitors. A site owner can opt in to a random token in browser storage, which recognises people across days and adds a returning-visitors card, with the consent question that comes with it. By default the overview shows visitors, visits, bounce rate and average visit next to the pageview total. There is no unique-pageviews figure: top pages are counted in pageviews. Lomen is free during the public beta.

Numbers you can ask questions

Lomen is one cookie-free script: sources, campaigns and AI referrals on one dashboard, and an AI that answers from your own figures. Free during the beta, no credit card.