The little "5 min read" label on an article is doing more work than it looks. It sets an expectation before anyone starts, and readers use it to decide whether to begin at all. Get it badly wrong and you either lose people who had the time, or annoy the ones who did not.
Where the Number Comes From
This calculator defaults to 238 words per minute, with Slow (150) and Fast (350) available alongside it. That middle figure comes from a 2019 meta-analysis by Marc Brysbaert, which pooled results across 190 studies covering more than 18,000 participants. For silent reading of non-fiction English prose, the mean landed at 238 words per minute, with most readers falling between 175 and 300.
Earlier tools often used 200, 250 or 265, numbers that circulated for decades without a clear source. The 238 figure is the best-supported single estimate we have, but a single estimate is exactly what it is. Reading speed varies with the reader, the material and the reason they are reading.
The same meta-analysis found reading aloud averages 183 words per minute, which is the number to use for scripts.
Realistic Speeds by Content Type
Use these as divisors when the default estimate feels wrong for your material.
| Content | Words per minute | Why it differs |
|---|---|---|
| Fiction, casual prose | 260-300 | Predictable syntax, familiar vocabulary |
| General non-fiction | 230-250 | The baseline the default is built on |
| Business and marketing copy | 200-230 | Scanned rather than read line by line |
| Technical documentation | 100-150 | Re-reading, cross-referencing, code samples |
| Academic papers | 80-120 | Dense citations, unfamiliar terminology |
| Legal text | 60-100 | Deliberate slow reading; every clause matters |
| Reading aloud | 150-183 | Constrained by speech, not comprehension |
| Second-language reading | 100-180 | Varies widely with fluency |
What a Word Count Cannot See
A reading time estimate divides words by a constant. Everything that is not a word is invisible to it.
Code blocks are the worst offender. A 40-line snippet might be 120 words but takes a developer two minutes to read properly, because reading code is not linear. Tables have the same problem: readers scan back and forth rather than reading left to right once.
Images, diagrams and charts add time that no word count captures. A single architecture diagram can hold a reader for a minute. Footnotes and citations pull attention away and back again.
The practical adjustment is to add roughly 10 to 15 seconds for each substantial image, table or code block, then compare that to what the calculator gave you. If the gap is large, your estimate is misleading and you should raise it manually rather than publish a number you know is wrong.
Timing a script for video or audio
For narration, divide by 150 rather than 238. Professional voice work sits between 140 and 160 words per minute for content meant to be understood, faster for advertising, slower for instruction. A 1,500-word script runs about 10 minutes spoken, not the 6 minutes a silent-reading estimate suggests.
Using Reading Time Well
The estimate is a promise to the reader. These habits keep the promise honest.
Round up, never down
Showing 7 minutes for something that takes 8 breaks trust. Showing 8 for something that takes 7 costs you nothing. Readers forgive finishing early.
Recalculate after editing
Most drafts lose 15 to 20 percent in the final pass. A label written before editing is usually stale by publication, and stale labels are worse than none.
Adjust for your audience, not the average
A blog for senior engineers can use a faster figure than one for beginners reading about the same subject. The material is identical; the readers are not.
Use it as an editing signal
If a how-to guide comes out at 22 minutes, the problem is usually structure rather than length. That is often two articles that have not been separated yet.
Estimate your reading time
Paste a draft and see the number update as you edit.
Does Showing Reading Time Actually Help?
Medium popularised the label in 2014 and reported that it increased the share of articles read to completion, though the underlying data was never published. Since then the practice has spread far enough that its absence is more noticeable than its presence.
The honest position is that reading time helps most where the reader has a real choice about when to start. On a long-form article opened from a newsletter, it lets someone defer to a better moment instead of bouncing. On a short reference page that someone landed on from search with a specific question, it adds nothing, and a 1 min read label can even signal that the page is thin.
Use it on essays and guides. Skip it on documentation and reference material, where nobody reads front to back anyway.
Frequently Asked Questions
Related Tools
Keep Reading
Word Count Targets: What Editors, Platforms and Google Expect
Two counters can disagree by five percent on the same paragraph. Here is how counting actually works, and the targets worth writing to.
Sentence Length and Readability: The Number Behind Every Score
Flesch-Kincaid, Gunning Fog and SMOG share one dominant input. What the research supports, and why counting sentences is harder than it looks.
Why Character Counts Disagree: Emoji, Accents and SMS Segments
One emoji can cost seven characters, and a single curly apostrophe can triple your SMS bill. The three definitions of a character, and when each applies.
Lorem Ipsum: The 2,000-Year-Old Text Hiding in Your Mockups
It is a scrambled passage of Cicero from 45 BC. Why fake Latin beats real English for layout work, and the design problems it quietly hides.