A reading-time estimate is one division: minutes = words ÷ WPM. The arithmetic is rarely the problem. Estimates go wrong because the speed does not match the reader, the word count does not match the text, or the result is treated as more exact than it is. The ten mistakes below fall into those three groups, and each comes with a fix you can apply in the reading time calculator.

Mistakes with the speed

1. Using a WPM that does not fit the audience

The calculator cannot know who will read your text. If you leave it on a default and your readers are not typical of that default, the estimate is wrong before it starts.

Take an 1,800-word onboarding document. At 238 WPM it comes out at 7 min 34 sec. But a new starter reading unfamiliar procedures is not reading at a general-interest pace. At a careful 180 WPM the same document is 10 min, so an eight-minute slot in an induction timetable will leave many people unfinished.

Fix: decide who the reader is before you choose a speed. Use the 238 WPM benchmark for a general adult audience reading ordinary non-fiction. Go lower for newcomers to a subject, people reading in a second language, or anyone who must act on the detail. When the audience is mixed, publish the more cautious figure.

2. Treating 238 WPM as everyone’s speed

The 238 WPM figure is an average from Brysbaert’s 2019 meta-analysis of 190 studies of English-speaking adults reading non-fiction. An average is the middle of a spread. The same paper gives a typical range of about 175 to 300 WPM for non-fiction and a higher average, 260 WPM, for fiction.

It is not a target, a pass mark or a description of you. A reader at 190 WPM and a reader at 280 WPM are both within the normal range, and the first needs almost half as long again for the same report.

Fix: use 238 WPM when estimating for people you do not know. For your own reading, use your own number. Our guide to average reading speed explains what the research figures do and do not tell you.

3. Confusing silent reading with speaking

The presets in a reading-time calculator are silent-reading speeds. Speech is slower. Brysbaert (2019) puts average oral reading at 183 WPM, and a talk delivered to an audience is often slower than plain reading aloud.

A 500-word announcement takes 2 min 6 sec to read silently at 238 WPM. Read aloud at 183 WPM it takes 2 min 44 sec, and with pauses for emphasis it will run longer.

Fix: for anything spoken, enter a custom WPM and then check it against a timed rehearsal. The details are in reading time versus speaking time.

4. Ignoring how difficult the text is

Word count measures length, not effort. A 2,500-word feature article and a 2,500-word contract both come out at 10 min 30 sec at 238 WPM. At 180 WPM the figure is 13 min 53 sec, and for close reading of legal wording even that may be optimistic.

Purpose matters as much as the text. Carver (1992) estimated typical rates for the college students he studied at about 300 WPM for ordinary reading, 200 for reading to learn and 138 for memorising. Those are estimates for his population, but they show how far the pace falls when the goal changes.

Fix: lower the speed for technical vocabulary, unfamiliar subjects and anything you need to remember or apply. Better still, time yourself on one page of the real document and use that rate. See why difficult text takes longer to read.

Mistakes with the count and the rounding

5. Rounding down, or rounding inconsistently

A 1,000-word article takes 4 min 12 sec at 238 WPM. Labelling it “4 min read” promises less time than the average reader needs. This site rounds labels up, so that article is a 5 min read.

Any whole-minute rule creates boundaries. Rounding to the nearest minute makes a 1,000-word post (4 min 12 sec) a “4 min” read and a 1,150-word post (4 min 50 sec) a “5 min” read. The two differ by 38 seconds but look a full minute apart. Rounding up has similar boundaries; its benefit is a more cautious estimate at the assumed pace. A separate problem is mixing rules across articles, which makes the labels harder to compare.

Fix: pick one rounding rule, write it down and apply it everywhere. Rounding up is the safer choice. Keep the exact figure for planning and use the rounded one only for labels.

6. Assuming every tool counts words the same way

Paste the same text into three tools and you may get three word counts. There is no single definition of a word. The Unicode Consortium’s segmentation standard (Unicode Standard Annex #29) describes default word boundaries and is explicit that they need tailoring for different languages and uses.

The disagreements come from a few places: hyphenated compounds, dashes set without spaces, numbers and contractions. Consider this sentence:

A well-known, state-of-the-art method—tested in 2024—cut costs by 3.5%.

Counting rule Count
Split on spaces only, so “method—tested” is one word 9
This site: dashes separate words, hyphenated compounds stay whole 11
Split on hyphens as well 15

This sentence was built to exaggerate the effect. In ordinary prose the gap is much smaller, and it matters most when you compare counts from two tools or work to a strict word limit.

Fix: use one tool for both the count and the estimate, and do not mix a word processor’s count with another tool’s timing. The rules this site uses are set out on the methodology page.

7. Counting navigation and boilerplate

If you select a whole web page and paste it, you bring along the menu, cookie notice, author bio, related links, comments and footer.

Suppose the article is 1,200 words and the surrounding material adds 400. At 238 WPM, 1,200 words is 5 min 3 sec, a 6 min read. With the extra 400 the count is 1,600 words, which is 6 min 43 sec, a 7 min read. The label is now a minute too long. The same thing happens with documents when you include the table of contents, running headers or a long reference list.

Fix: paste only the body text a reader is expected to read. Copy from the editor or the source file instead of the published page where you can, and check that the word count is close to what you expect before trusting the time.

8. Ignoring content that is not prose

A calculator counts words and nothing else. It gives no time to an image, a chart or a diagram, even though a reader may study one for a while. Code and tables fit badly too: a short code block can take far longer to understand than a paragraph of the same size, and a data table is not read left to right at 238 WPM.

Take a tutorial with 1,500 words of explanation and a dozen code samples. The prose alone is 6 min 18 sec at 238 WPM. Anyone who works through the samples will spend much longer.

Fix: estimate the prose with the calculator, then add your own allowance for the rest, ideally by timing yourself on one representative piece. A label such as “7 min read, plus time for the exercises” is more useful than a single number that ignores them.

Mistakes with interpreting the result

9. Treating the estimate as exact

The calculator will tell you that 1,800 words takes 7 min 34 sec. The seconds are a product of the division, not a measurement of anything. The inputs are approximate: the speed is an average or a guess, and the count depends on the counting rule.

A more truthful reading of that result is a range. The same 1,800 words is 6 min at 300 WPM and 10 min at 180 WPM, and where a given reader lands depends on the reader, the text and the day.

Fix: round for communication and plan with a margin. For a schedule, use the slower end of the range. For a published label, a whole number of minutes is as much precision as the method supports.

10. Using one speed-test result as your permanent pace

A single run of the reading speed test tells you how fast you read that passage, on that screen, at that moment. It will shift with the topic, your tiredness, the time of day and how carefully you chose to read. Measured speed also depends on the test text itself. Trauzettel-Klosinski and Dietz (2012) built standardised passages in 17 languages for exactly this reason: without carefully matched texts, results are hard to compare.

If three attempts give you 221, 247 and 236 WPM, none of them is your true speed. Together they suggest a pace in the mid-230s for that kind of text. A fast result that came with poor comprehension is not a pace you can plan with at all.

Fix: take several measurements on different days, use the middle value, and re-measure every so often. Keep separate figures for easy and demanding material if you read both. Our guide to building a personal reading speed baseline sets out a simple routine.

Before you rely on a number

Check three things: that the pasted text is only what will be read, that the speed matches this reader and this purpose, and that you have rounded up and left a margin. An estimate that passes those checks is good enough to label an article or budget a meeting, provided you remember that it describes a typical reader and not a particular one.