A visitor counter goes up when the snippet runs. On MapMyVisitors that snippet is an image or a small script. Each time a browser requests it, the count increases and, if you use the map, a place is stored with that request. The number is loads of the widget. It is not a census of people.
That sounds pedantic until you quote the number. "10,000 visitors" and "10,000 widget loads" are different claims. Only the second one is what the counter measured.
The unit, in practice
One person who opens three pages with the widget in the template produces three counts, if each page requests it. One person who opens the same page twice produces two, unless a cache served the page without requesting the image again. A bot that fetches the snippet produces a count too.
A page that is cached without the widget produces nothing. The article was read and the counter did not move. That is the tradeoff of counting the snippet rather than parsing your server log. It is also why the install check matters: if the request is not leaving the browser, the counter is silent. Checking that the widget is recording is that test.
Reloads, prefetch, and link previews add loads no one would call a visit in conversation. You cannot remove all of them after the fact. You can avoid pretending the integer is a headcount.
What to say instead
Say "the widget loaded about 10,000 times this month" if you need to be exact. Say "on the order of a few thousand visits" if you are describing the site to a person and you have looked at the log enough to know it is not mostly bots. Do not say "10,000 unique readers." Uniqueness is a different metric, and this counter does not compute it. The distinction is the whole of what a unique visitor is.
A pageview in a full analytics suite is a close cousin: also a load, also not a person, but defined by that product's tracker and its idea of a page. Your server logs are a third cousin. The three numbers will not match. Quote one of them and name it.
When the count is still the right tool
You want a public number or a public map, you want it without a tag manager, and you want a rough sense of whether the site is being fetched. A counter does that job well. It does not do sessions, funnels, or "how many humans." If that is the question, you want a different product, and a map versus a hit counter is the short comparison of the two simpler tools before you jump to a suite.
