How Do You Know If People Actually Read Your Content?

The honest answer, for most sites, is that they don't. Time on page and bounce rate are the two numbers editors reach for, and both are proxies for reading rather than measurements of it.

Time on page counts how long a tab was open, not how much of that time was spent reading. A reader who opened an article, got distracted, and came back to close the tab twenty minutes later looks identical in the data to someone who read every word slowly. Bounce rate has the opposite problem: it can look great on a page nobody actually read, if they didn't click anywhere else before leaving.

What would actually answer the question

To know whether someone read an article, you need a measurement tied to their position within the content itself, not the duration of their visit or whether they navigated elsewhere afterward. That's the difference between a session-level proxy and a content-level measurement.

Reader Progress does this by watching four fixed points in a reader's path through a piece: Started Reading, Reached the Middle, Read Most, Reached the End. Detection is based on where the reader's viewport sits relative to the content, not a timer and not a click. A milestone firing means the reader's position genuinely reached that point in the piece.

What this looks like on a real article

Say an article's Started Reading count is strong but Reached the Middle drops sharply. That's not a hunch, it's a specific, countable signal that most readers are leaving at a specific point, usually because something in the opening third of the piece isn't holding attention. Time on page would have shown you an average duration with no way to know where in the piece that average time was actually spent.

This is the core idea behind content engagement analytics as a category: measuring the content experience directly, rather than inferring it from visit-level proxies that were built to answer a different question.