Show a “min read” label if your posts are long-form prose that people read from top to bottom, and if you are willing to calculate it the same way every time. Leave it off very short posts, reference pages and anything people use instead of read. A label is a statement about length, and it is only worth making when the length is something the reader needs help judging.
One caveat before the details. You will see claims that reading time labels increase engagement or improve search rankings. None of the research this site relies on tests either claim, so we do not make them. Evidence that a label changes reader behaviour is limited, and we cannot support any statement that it is a ranking factor. The case for a label below rests on something more modest: it tells readers a true and useful thing before they commit.
What a label tells the reader
A person who lands on an article cannot see how long it is. The scroll bar gives a rough hint on a desktop and almost none on a phone. A label answers the question the reader would otherwise answer by scrolling to the bottom and back: is this a two-minute thing or a twenty-minute thing, and do I have time for it now?
That is a planning question, and it is why the label matters most where lengths vary. On a site where every post runs about 600 words, readers learn the length quickly and a label repeats what they know. On a site that mixes 500-word notes with 5,000-word essays, it separates the two before anyone clicks.
It helps to be clear about what the label does not mean. Nielsen (1997) reported that most web users scan pages instead of reading them word by word. Many visitors to a “9 min read” will spend one minute on it, picking out a heading and two paragraphs. The label does not predict that behaviour. It describes how long the piece takes if read in full at an assumed pace, and that is how it should be understood: a measure of size expressed in minutes, not a forecast of time on page.
When a label helps and when it does not
| Page type | Label? | Reason |
|---|---|---|
| Essays, features, long tutorials in prose | Yes | Read linearly; length varies; readers plan around it |
| News and opinion of mixed lengths | Usually | Helps readers choose between a brief and a long read |
| Posts under about 300 words | Usually unnecessary | At 238 WPM, labels are only one or two minutes, so they distinguish little |
| Recipes | No | The useful times are preparation and cooking, not reading |
| Reference pages, documentation, glossaries | No | People look things up; nobody reads them through |
| Pages that are mostly images, video or data | No | Word count is a poor measure of the content |
| Code-heavy tutorials | Maybe | Only with a stated rule for how code is treated |
The short-post case is worth a number. A 150-word update takes 38 sec at 238 WPM. The label rounds up to “1 min read”, as it would for a 40-word post or a 230-word one. When every item in a list carries the same label, the label has stopped carrying information. Hide it below a threshold of your choosing; 300 words is a reasonable house rule, not a research finding.
Recipes show the other failure. A recipe page might contain 900 words, most of it an ingredient list and numbered steps that the cook will read in pieces over forty minutes with flour on their hands. “4 min read” is arithmetically correct and practically meaningless, and it sits awkwardly beside the preparation time the reader came for.
Choosing a method
A reading time label is the output of a small set of decisions. Make them once, on purpose.
The reading speed
The speed assumption changes the label more than anything else. Take a 2,400-word post:
- at 180 WPM it is 13 min 20 sec, labelled “14 min read”;
- at 238 WPM it is 10 min 5 sec, labelled “11 min read”;
- at 300 WPM it is exactly 8 minutes, labelled “8 min read”.
Same article, three different promises. For a general audience reading English non-fiction, 238 WPM has the best support: it is the average from the meta-analysis by Brysbaert (2019), which also notes that the commonly repeated figure of 300 WPM is higher than the evidence warrants. A site with a specialist audience and dense material has good reason to choose something lower. The background is in our guide to average reading speed.
Whatever you choose, it describes a group. Individual readers range widely on either side, so the label will be generous for some and tight for others. That is unavoidable and worth saying somewhere on the site.
The rounding
Round up to the next whole minute, with a minimum of one. A post that takes 10 min 5 sec is not a ten-minute read for anyone at or below the assumed pace; it is a little more. Rounding up makes the label an upper bound for the reader you assumed, and nobody complains about finishing early. The arithmetic and its edge cases are set out in how reading time is calculated.
What gets counted
This is where sites drift apart, usually without noticing. Decide each of these explicitly:
- Body text and headings: counted. This is the article.
- Code blocks: counted or excluded, but pick one. Consider a tutorial with 1,100 words of explanation and code samples that a word counter sees as another 400. Counting everything gives 6 min 18 sec and a “7 min read”. Counting prose alone gives 4 min 37 sec and a “5 min read”. Neither is correct in any deep sense, because people do not read code at prose speed; some skip it and some study it line by line. What matters is that the rule is the same on every tutorial.
- Tables: the cell text can be counted as words, though a reader comparing figures across a table will take longer than the count implies.
- Images: they contain no words, so they contribute nothing unless you add a fixed allowance per image. Any such allowance is a house rule of thumb. Say so if you use one.
- Captions, footnotes, pull quotes, author bios, navigation, comments: generally excluded. Pull quotes repeat body text and would be counted twice.
Then write the method down in one sentence, for example: “Reading times assume 238 words per minute, count body text and headings only, and are rounded up to the next minute.” If you cannot state your method that briefly, it is probably being applied inconsistently. A step-by-step version of this process for a single piece is in how to estimate reading time before publishing.
Consistency across the site
A label is only useful relative to other labels. Readers calibrate: after a few visits they learn that your “6 min read” takes them four minutes, or eight, and they adjust. That calibration works only if every label on the site came from the same rule.
The usual ways it breaks:
- The content system computes labels for new posts, while older posts carry numbers someone typed by hand.
- A redesign changes the speed assumption and only newly published posts pick it up.
- Authors can override the figure, and some shorten it to make a long piece look more approachable.
- The label is calculated from the page as rendered, so a post with a long comment thread or a large related-links block reads as longer than it is.
The fix is dull and effective: one function, one speed, one rounding rule, applied to the same part of every page, with no manual override. If you change the method, recalculate everything. To check that your system’s output is what you expect, paste a post’s body text into the reading time calculator with your chosen speed and compare the result with the label on the page.
Keeping the label honest
The temptation with any number shown to readers is to tune it. A shorter label looks friendlier, so the speed assumption creeps up to 300 or beyond, or the rounding flips to round down. Each choice shaves a minute and makes the label a little less true.
The reader notices eventually. Someone who is told five minutes and needs eight will not time it, but they will learn that your labels run short and discount them. At that point the label is decoration.
A few habits keep it truthful:
- Use a speed you can justify from a published average or from knowledge of your audience, and do not pick it for the result it produces.
- Do not use labels to imply that a skim is a read. Rayner et al. (2016) are clear that skimming trades away detail; a label based on a skimming pace describes a different activity from the one the word “read” suggests.
- Leave the label off pages where it would mislead, even if the template makes it easy to show everywhere.
- Say “about” somewhere, in the label itself or in a short note on how times are calculated. It is an estimate, and readers are entitled to know the assumptions behind it.
Other pitfalls, on the calculation side, are collected in common reading time calculator mistakes.
So: show the label where length varies and the content is meant to be read through, compute it by a rule you can state in a sentence, apply that rule everywhere, and resist the urge to flatter the number.