This scans a block of text and ranks the words that show up most often, after filtering out common connector words like the and of that carry little topical meaning on their own. What's left is a quick, ranked list of what a piece of writing is actually about, useful for tagging, SEO, or just getting the gist of a long document fast.
How to use it
- Paste the text you want to analyze.
- Set how many top keywords you want.
- Read the ranked list that appears right away, each tagged with how many times it occurred.
- Edit the text and watch the ranking update live.
- Copy any keyword for use as a tag or a search term.
When it helps
- Generating a starting set of tags for a blog post based on its actual content.
- Getting a fast sense of what a long document is mainly about without reading all of it.
- Checking whether an article's most frequent terms match the keywords you were writing toward.
- Comparing the dominant themes across a few pieces of writing.
- Building a simple tag list from a batch of feedback or survey responses.
Good to know
- Analyze a longer piece of text for more reliable results. Something very short won't have enough repetition for a meaningful ranking.
- Increase the keyword count if the top few feel too generic, since less frequent terms are sometimes more specific.
- Pair this with the sentiment analyzer for a fuller quick read on a piece of text, what it's about and how it's said.
- Run the same text through before and after an edit to see whether your revisions shifted the piece's apparent focus in the direction you intended.
- If two pieces of writing are supposed to cover the same topic, run both through and compare their top keywords to check they're actually aligned.
Questions
Does this actually understand which words matter, or just count frequency?
It ranks by frequency after removing a fixed list of common stopwords, which works well as an approximation but doesn't understand meaning the way a full language model would.
Can it extract phrases instead of single words?
This version ranks individual words. Multi word phrase extraction would need a more advanced technique looking at word sequences rather than single word frequency.
Why do short or common words never show up?
Anything under three letters and a fixed list of connector words are filtered out before ranking, since they'd otherwise dominate almost any English text without adding real information.
Does capitalisation affect the count?
No, everything gets lowercased before counting, so Product and product count together as the same keyword.
Can I use the extracted words directly as SEO meta tags?
They're a useful starting point for spotting a page's actual topical focus, but meta tags should still be chosen deliberately based on search intent, not generated automatically from frequency alone.
Does punctuation affect which words get extracted?
Punctuation is stripped out before counting, so a word followed by a comma or a period is still counted correctly as that word.