Use the free Extract Keywords from Text below: your data never leaves your browser — all work happens on your device, and nothing is uploaded to any server.
Finding what a text is really about
Open a long article, report or transcript and ask a simple question: what is this about? The honest answer is usually a handful of terms that keep coming back. This tool finds those terms for you. It counts the words in your text, drops the common filler words that carry no topic, and ranks the rest by frequency, so the main terms rise to the top of a table you can read at a glance.
That is useful far beyond curiosity. Writers check whether a draft actually emphasises the topic they meant to cover. Students turn a chapter into a revision list. Researchers skim interview transcripts for recurring themes. Marketers compare the language of competing pages. In each case the keyword table is a fast first map of a text, not a replacement for reading it.
How the ranking works
The tool lowercases everything, strips punctuation, and splits the text into words. Words shorter than your chosen minimum length are dropped, and so is a built-in list of stopwords: the, of, and, to, in, is, you, that, and dozens more that appear in almost every English text without saying anything about its topic. The remaining words are counted, sorted by count with ties broken alphabetically, and shown as a table with each word and its count.
Two controls shape the result. “Show top” sets how many keywords appear, from five to one hundred. “Minimum word length” filters out short words, which is handy when two-letter abbreviations flood the table. Raise it if the table is noisy, lower it if a short term such as “AI” or “oil” is exactly what you are tracking.
What frequency cannot tell you
A word that appears often is not automatically the most important word. Introductions repeat background terms, authors repeat a word out of habit, and a crucial idea may appear once in a single sharp sentence. The tool also does no stemming, so “run”, “runs” and “running” count separately, and it treats phrases as single words only, so “machine learning” appears as “machine” and “learning” rather than one term. Read the table as a hint about emphasis, then judge importance yourself.
Getting better results
Longer texts give better tables: a few hundred words at least, ideally more. Clean the text first if it contains navigation menus, footers or repeated boilerplate, since repeated boilerplate becomes fake keywords. Compare the top ten against your own summary of the text; if they disagree, the disagreement itself is informative, because it shows the gap between what the text emphasises and what you thought it emphasised.
Practical uses
Bloggers check a draft against the search phrase they target. Teachers show students how to summarise a chapter. Editors spot when a manuscript drifts off topic. Job seekers compare a job description’s repeated terms with their resume. Analysts scan open survey answers for recurring words before reading the answers in full. The pattern is always the same: count first, think second, and let the counts guide where you look.
Keywords, keyphrases and search intent
Single words are only the first layer. In search and in writing, people think in phrases: someone searching for help does not type “tent” alone, they type something like “how to clean a canvas tent” or “best tent for rain”. A keyword table gives you the building blocks, and the next step is to read the original text around the top words and note which phrases they belong to. That combination, counts plus context, is what turns a mechanical table into an editorial decision about what to emphasise, what to cut, and which phrases deserve their own heading. It also explains why two texts with similar keyword tables can serve very different readers: the same words arranged around different questions are different articles.
Privacy note
All counting happens in your browser. Your text is never uploaded, stored or seen by anyone else, which matters when the text is unpublished work, private feedback or interview material. Paste, read the table, copy what you need, and close the page: nothing remains behind.