Word Frequency Counter: Practical Tips and Common Mistakes
Reviewed by the FreeOnline.fyi team · Updated 2026-09-16
What the counter actually tallies
The box on the Word Frequency Counter page works as you type: paste an article, essay or transcript and a ranked table appears within a fraction of a second, with no submit button involved. Underneath the textarea a small strip reports total words, unique words and characters, which is often the quickest sanity check that you pasted the whole document and not just the first screen of it.
Words are split on whitespace and punctuation, and in our testing capitalised and lower-case versions of the same term land in the same row — run a two-second test pasting "The the THE" if you want to confirm that before trusting a large number. Ties are broken alphabetically, so the ordering stays stable when several words share a count.
Two things the table will not do. It never shows more than the top 20 words, and it does not group related forms: "review", "reviews" and "reviewed" are three separate entries. That matters when you are judging whether a concept is genuinely repeated or merely inflected a few times.
Percentages beat raw counts
The percentage column is where the useful signals live. In a 500-word draft, a word appearing 12 times accounts for 2.4% of everything you wrote; the same 12 occurrences inside a 4,000-word report is 0.3% and barely audible to a reader. Counts look impressive, shares tell you whether anyone will actually notice.
Adding up the percentages of the top few rows is a fast measure of concentration. A draft where a handful of terms soak up a large chunk of the total tends to read as repetitive, while one where nothing exceeds 1% can read as vague, because no single idea is being driven home. That lopsided shape is roughly what Zipf's law predicts for natural language, so an unusually even spread is worth a second look rather than a pat on the back.
Keep an eye on the denominator too. The total stays the same when you flip the common-word toggle, because stopwords are filtered out of the ranking rather than out of the count. That means a word's percentage stays comparable across both views, which is handy when you switch mid-edit.
When to hide common words
With the toggle off, function words dominate: "the", "and" and "of" occupy the top rows of almost any English text. That is exactly what you want if you are studying sentence rhythm, or checking how heavily a writer leans on connectives and hedging words.
Switch the built-in English stopword list on and the content words underneath surface instead. For SEO and editorial work this is the useful view: it answers the practical question "what is this page actually about?" without opening a keyword tool, and a glance at the top rows against your intended topic is a fair check for drift. If the piece was meant to be about invoicing but "payroll" keeps appearing, something slid. Work on a meta description benefits from the same read, since the repeated terms are the honest candidates for a summary line.
The mistake is treating these percentages as a target. Search systems read meaning, context and links far more than they count tokens; a page with "concrete repair" at exactly 2% is not better than one at 1.2%. Use the numbers to spot unintended repetition and to confirm your main subject is present at all, then edit for the reader.
Exporting the table safely
The Download CSV button builds word-frequency-top20.csv entirely in the browser and stays disabled until at least one word has been found. Output follows the CSV convention described in RFC 4180, with values quoted and escaped so a stray comma or quotation mark inside a word cannot break the columns.
Spreadsheet quirks are worth knowing before you sort. Some applications interpret the percentage column as a date or a plain decimal on import; if the figures look strange, import that column as text and convert afterwards. Sorting by count rather than by share can also reorder rows surprisingly, since the two columns are proportional to each other within a single run.
The real value of the export is comparison over time. Keep one CSV per draft and diff the top rows after a rewrite: terms that dropped off the list tell you what you successfully trimmed, while new arrivals show what you accidentally emphasised. A single file is a snapshot; two files side by side are an editing log.
Mistakes that skew the numbers
The most common one we see is pasting boilerplate. Copy a web page from top to bottom and navigation labels, sidebars, cookie notices and footer links all get counted — and because those strings repeat across a site, they can outrank the words in the article itself. Paste the body text only.
Transcripts are the second trap. Filler such as "um", "like" and "you know", plus speaker names and timestamps, will crowd out the substance; strip the speaker labels first if you want a clean signal. Dialogue-heavy fiction has the same shape, since character names repeat for good reasons.
Then there is over-correcting. Chasing a lower top percentage by replacing every second mention of a key phrase with a synonym can make a piece harder to follow than the repetition it was meant to fix. Repetition is a problem when it is accidental, not when it is the subject you are writing about — and a page that keeps its main term steady is usually easier to skim than one that keeps renaming things.
Privacy, limits and honest caveats
All analysis happens locally in the page, so nothing you paste is uploaded. That is why the tool is safe to point at unpublished drafts, client documents or interview transcripts. It handles roughly two million characters comfortably, and results refresh with a light delay after you stop typing so large pastes stay responsive.
It is a frequency tool, not a grammar checker, plagiarism detector or readability score. It will not tell you whether a sentence works, and it only knows the English stopword list built into it — other languages will show every word, including their equivalents of "the". For most non-English text, leave the toggle off and read the raw ranking.
If you need a quick look, keep this page open while you write. For the same treatment of other text chores, the rest of our free tools run the same way, in the browser and without an account. And as with any automated count, spot-check an important figure by hand before you quote it in a report or a client deck.