The existence of redirects can cause some confusion over the data when it comes to page views, since it appears that the view only gets counted for the page title of the redirect, even though, in practice, a redirect means that a visitor *views* the target page. There are valid reasons in MediaWiki for the redirect page and the target page to be treated separately, but for many metrics-related use cases, you probably actually want to know the aggregate counts for the page you are querying as well as all redirects to it—or at least have the option to include them with a togglable parameter.
- I have a list of articles compiled from some source, like a category, contributions list, or WhatLinksHere. One of the articles in this list was recently renamed, so even though I have the current name from the source list, most of its views are counted under the former page name, which is just a redirect I don't know about. In this case, if the current name didn't exist at all within the period I am querying, even though it's a longstanding article that did exist under a previous name, my query gives me an unhelpful 404.
- I am querying for pageviews because I want to measure the popularity of a particular article, in order to compare it to others. My data won't be as accurate as intended when some articles in the set have very canonical names ("Canada," "Mexico"), while others may be known by several common names redirecting to one article title ("USA" and "United States of America" redirect to "United States"). Sometimes, for Manual of Style reasons, Wikipedia even chooses lesser-used but more correct names for titles. In this case, I would have wanted the option to include all redirects' page views as well, since including "USA" and "United States of America" page views gives me a better sense of the total viewership of "United States" compared to other articles.
In fact, for analytics, it probably makes most sense for redirects to be included by default—with the option, of course, to exclude them if you legitimately want to know the page views of a page without redirects, but that seems rarer. Especially since page moves in an article's history can really mess with your data in unexpected ways.