Reading time calculator

How fast people actually read

There is no single reading speed, and the honest ranges are wide enough to triple the answer. What people usually mean by "reading speed" is silent reading of ordinary prose, in a first language, with nothing at stake. That is one situation out of many.

The ranges below are the ones commonly quoted for adults. Treat them as brackets rather than measurements: a reader moves around inside them depending on the material, the hour of the day and how much the text matters to them.

What slows a reader down

Almost everything that pushes the real rate below the average lives in the reader or the screen, not in your prose. You cannot fix most of it. You can stop pretending it is not happening when you pick a rate.

The heaviest factor is language. Someone who reads a second language well still reads it more slowly than their first, and the gap widens as the vocabulary gets less common. If a large share of your audience learned your language later, the average rate is the wrong rate.

Density comes next. Unfamiliar terms send the eye backwards: the reader reaches the end of a sentence, realises they lost the thread at the third clause, and starts again. That backtracking appears in no word count anywhere, and it is most of the distance between 250 words per minute and 130.

Why two sites label the same article differently

Put an identical article through three publishing platforms and you can get three different labels. Nothing is broken. The label is a design decision each platform made once and then applied to everything.

Three things differ between them. The first is the assumed rate: 200, 225 and 250 words per minute are all in circulation, and a 1,000-word article comes out at five minutes, four and a half, or four, depending on which figure is baked in. The second is what got counted, because some tools measure the article body alone while others fold in headings, image captions, the author biography and the footnotes. The third is rounding, which almost always goes up to the next whole minute and never below one.

Some platforms also add a fixed number of seconds per image. That is a defensible thing to do and an invisible one, because the label never mentions it.

So a published reading time is comparable to other labels from the same source and to very little else. If you are putting a number on your own work, the move that helps is naming the rate beside it.

What a word count cannot see

A word counter counts words. It has no idea that the paragraph you just wrote is followed by a forty-row table, a screenshot with eight numbered callouts, and a code block the reader is meant to run.

Those elements do not slow reading proportionally. They interrupt it. Nobody reads a table: the reader finds the row that concerns them, reads across, and leaves. A code block gets read line by line at something close to study pace, then read again. A footnote charges for the jump, the note, and the effort of finding the sentence you were in.

No formula recovers this, because the cost depends on what the element is for. A decorative photograph costs a second. A diagram that carries the argument costs as long as the paragraph explaining it. You know which one you published; the calculator does not.

• Decide once what each kind of element is worth, then apply it everywhere, so your own labels stay comparable with each other.
• A workable starting set: a couple of seconds for a decorative image, ten or more for a figure the reader has to interpret.
• Price a code block at the careful rate rather than the prose rate, and add more if the reader is expected to run it.
• For a reference table, count the header row plus two data rows. That is roughly what anyone consumes.
• For footnotes, add the notes themselves plus a few seconds each for the round trip.
• These are judgement calls, not measurements. Write yours down, keep them stable, and say what they are if anyone asks.

Pick the pace for the person actually reading

The middle preset assumes an adult reading prose in a first language, once, with nothing depending on it. Change any part of that and the number should change with it.

When the choice is close, err slow. A reader who finishes ahead of your estimate feels efficient. A reader still going at twice the stated time decides the number was marketing.

Three questions settle the rest faster than any table:

• Who is reading, and in which language? A team spread across four countries is not a general audience.
• What will they do with it — read it once, act on it, or hold it for an exam? Those are three different rates.
• What does being wrong cost? A courtesy label on a newsletter can be roughly right. A weekly reading load you have just assigned to a class cannot.

Where the number earns its place

A reading time pays for itself where someone is deciding whether to start now or later. On a newsletter or a long post it answers that at a glance, which is the entire job.

It does a second, quieter job for the writer. Documentation that has drifted past fifteen minutes usually wants splitting, and the estimate notices the drift before any reader complains. A pre-read circulated with an honest twelve minutes on the front tends to get read; the same document with no number on it gets opened during the meeting.

It is worth least on pages people arrive at with one specific question. Someone hunting a single command in a reference page was never going to read the page, and telling them it takes nine minutes says nothing useful about their afternoon.