Word Counter: How It Works
A word counter answers a question with more edge cases than it appears to have. Hyphenated compounds, numbers, contractions and headings are all counted differently by different tools — which is why your count and your editor's rarely agree exactly.
How words are counted here
Text is split on whitespace, and any resulting token containing at least one letter or digit counts as a word. That means:
| Input | Counted as |
|---|---|
state-of-the-art | 1 word |
don't | 1 word |
2026 | 1 word |
— (dash alone) | 0 words |
e.g. | 1 word |
Word processors apply similar rules, but not identical ones — some count a hyphenated compound as multiple words, and most exclude footnotes, headers and captions from the main count. Differences of one or two percent between tools are normal and rarely matter.
Where the count actually counts
| Context | Typical target |
|---|---|
| Blog post | 800 – 2,000 |
| In-depth guide | 2,000 – 5,000 |
| Academic essay | Specified, usually ±10% tolerance |
| Press release | 300 – 500 |
| Executive summary | Under 500 |
| Cover letter | 250 – 400 |
Academic limits are the ones worth taking literally. Many institutions apply a penalty above the stated tolerance, and their counting rules — whether footnotes, references, tables and appendices are included — are specified in the handbook and vary between departments. Check before trusting any tool's total.
Reading time
Adult silent reading averages 200–250 words per minute for general prose, and slower for technical material. Dividing the word count by 225 gives a serviceable estimate, and this is roughly what publishing platforms display. Treat it as an indication of scale rather than a measurement: readers skim, re-read and stop, and none of that is in the arithmetic.
Length and search performance
Longer pages correlate with better rankings, and the correlation is widely misread as causation. Comprehensive content tends to be longer because covering a subject properly takes words — not because length itself is rewarded. Padding a 600-word answer to 2,000 words produces a worse page that performs worse. The useful question is whether anything a reader needs is missing, not whether a target has been reached.
On writing shorter
Most first drafts lose 10–20% without losing meaning. The reliable cuts are adverbs that add nothing, 'in order to' for 'to', 'due to the fact that' for 'because', hedging stacks like 'it may perhaps be possible', and opening sentences that clear the throat before the point. Reading a draft aloud finds these faster than any tool.