Reading Time, Explained: How to Turn Word Count Into Honest Minutes

Every blog post seems to carry a little badge near the title: "6 min read." Most of those numbers are guesses, or worse, hard-coded into a template years ago and never touched again. I have shipped posts where the CMS swore the article was a "3 min read" while it actually took a careful reader closer to nine minutes. The gap matters, because a reader who budgets three minutes and hits minute eight feels misled, and the back button is one click away.

So let me walk through what "reading time" actually means, how the minutes get calculated, and where the same math quietly shows up in writing speeches and short-video scripts.

How fast adults actually read

The single number that anchors everything is reading speed, measured in words per minute (wpm). The casually repeated figure is "200 wpm," and that is a fine rough anchor for adult silent reading. But the most defensible number comes from Marc Brysbaert's 2019 meta-analysis, which pooled more than 190 studies and landed on a silent-reading mean of about 238 wpm for non-fiction English prose. That is the figure the Reading Time Calculator uses as its "Average" preset, precisely because it is sourced rather than folklore.

Real readers spread out around that mean. Dense technical or legal text, the kind you have to parse one clause at a time, drops people toward 150 wpm. Familiar, skimmable content pushes past 300 wpm. So a single fixed wpm for every article is already a simplification — the honest move is to pick a band that matches the material:

• Slow (150 wpm): contracts, API references, study material a student must absorb.
• Average (238 wpm): ordinary blog posts, news, documentation prose.
• Fast (300 wpm): lists, FAQs, content the audience already half-knows.

Words to minutes: the actual arithmetic

The conversion is plain division: minutes equal word count divided by wpm. Nothing fancy. The craft is in counting the words correctly and presenting the result in a way that does not lie by rounding.

Here is a concrete example. Suppose you paste a 1,500-word how-to article and leave the speed on Average:

1,500 words ÷ 238 wpm = 6.30 minutes
0.30 minutes × 60 = ~18 seconds
→ about 6 min 18 seconds, which a badge rounds to "6 min read"

That 6.3 minutes is your honest estimate. If you had used the lazy "200 wpm" default instead, you would have shown 7.5 minutes — a full minute and a half longer than reality, which makes the article look more daunting than it is. Small wpm differences compound across thousands of words, so picking a sourced number is not pedantry; it changes the badge.

A quick sanity check on word count itself is worth doing before you trust any timing tool. If you want to confirm how many words and characters are really in a draft, a dedicated word counter gives you the raw count, and a word frequency counter shows which terms you lean on too hard. Reading time is only as accurate as the count feeding it.

Why Chinese (and Japanese and Korean) break the formula

The word-divided-by-wpm formula assumes words are separated by spaces. English obliges. CJK scripts do not. Chinese, Japanese, and Korean run characters together with no whitespace boundaries, and a fluent reader paces through them character by character, not "word" by "word."

Drop a 2,000-character Chinese essay into an English-only counter and it might report a few dozen "words" — a number so wrong it is useless. The fix is to count CJK per character and scale on a characters-per-minute band instead. Roughly, the three speed presets map to about 260, 360, and 500 characters per minute. A mixed line like AI 模型 GPT counts as two Latin words (AI, GPT) plus two CJK characters (模, 型): four reading units, each paced on its own clock. Any tool that does not split mixed scripts this way will misreport bilingual content, which is one of the most common reading-time mistakes I see.

The "X min read" badge, done right

The badge at the top of a post is just a reading-time estimate formatted for humans. The trouble is that most CMS implementations bake in three errors at once: a fixed wpm, no mixed-script handling, and a value that was correct when the draft was 800 words but never updated after the final edit ballooned it to 1,400.

The reliable workflow is boring and it works. Finish the post. Paste the final body into the calculator. Read off the real number. Write that number into your front-matter so every template that renders the badge stays honest. When I started doing this, the gap between "claimed" and "actual" read time on my own posts collapsed from minutes to seconds, and that is exactly the kind of small trust signal that keeps a reader from bouncing two-thirds of the way down.

Speaking time is a different number entirely

Here is the trap that catches more people than the badge: silent reading and reading aloud are not the same speed, and confusing them is how talks run over time.

Silent reading sits at 200 to 300-plus wpm. A clear, audience-friendly speaking pace is far slower — around 130 wpm, which is roughly what TED targets (about 130 to 150). So the same 1,000-word passage that reads silently in about four minutes takes nearly eight minutes to deliver out loud. Plan a five-minute talk slot off the silent-reading estimate and you will run roughly double.

To work backward from a time slot, multiply by ~130 spoken words per minute:

• 5-minute talk: about 650 words
• 10-minute talk: about 1,300 words
• 20-minute conference slot: about 2,600 words

For Mandarin, figure ~200 characters per minute, so a five-minute talk is roughly 1,000 characters. The same logic caps a short-video script: 60 seconds of speech is about 130 English words. Paste a draft, read the speaking-time box, and trim or pad until it fits the slot — long before you book studio time and discover the overrun in the recording.

If you are curious how your own pace stacks up, a typing speed test is a fun companion: typing wpm and reading wpm both describe how fast you move through text, just through different muscles.

Putting it together

Reading time is one short formula — word count over wpm — wrapped in three judgment calls: pick a sourced reading speed (238 wpm is the defensible default), count mixed scripts the way they are actually read, and never reuse a silent-reading number to plan something spoken. Get those three right and your badges stop lying, your talks stop running over, and your scripts fit their format on the first take. Paste your text, pick the speed that matches the material, and read off two honest numbers instead of one hopeful guess.

---

Made by Toolora · Updated 2026-06-13