Word Cloud Generator

How to use this tool

1. 1

   Paste your text

   An article, a set of reviews, survey responses, support tickets, meeting notes — anything where you want to see which words dominate.

2. 2

   Filter the noise

   "Ignore common words" removes the, and, of and similar. Add your own exclusions for words that are technically frequent but tell you nothing — your brand name, for instance.

3. 3

   Tune the sizes and word count

   Fewer words at a larger size reads better on a slide; more words at a smaller size shows more of the distribution. The minimum-appearances slider hides one-off words.

4. 4

   Pick colours and rearrange

   Five palettes and a custom background. Rearrange reshuffles the layout without changing the data, which sometimes fits more words in.

5. 5

   Download PNG or SVG

   PNG for slides and documents, at double resolution so it stays sharp. SVG for print or any use where it needs to scale without blurring.

What word clouds are good at

They are good at one specific thing: showing you, quickly, what a body of text is mostly about. Dropped into a hundred customer reviews with no prior expectation, a word cloud will surface the recurring subject in seconds.

That makes them genuinely useful as a first pass — an orientation step before proper analysis, or a way to check whether a piece of content is about what you intended it to be about.

They are also effective in presentations for the same reason. An audience takes in a word cloud in a moment, whereas a frequency table takes real attention. As a way of saying "this is the shape of what people said", it works.

What they are bad at, and why it matters

Word clouds have well-known limitations, and knowing them keeps you from over-reading one. They discard word order entirely, so "not helpful" and "very helpful" both contribute to a large "helpful".

They also discard context. A word appearing frequently tells you it was said often, not whether it was praised or complained about. Sentiment is invisible.

And the visual encoding is imprecise: readers judge size by area, but longer words occupy more space at the same font size, so a long word can look more important than a short one with a higher count. This is exactly why the frequency table sits below the cloud here — the picture is the summary, the numbers are the evidence.

Why filtering changes everything

Without stop-word filtering, every word cloud of English text looks identical: "the", "and", "of" and "to" dominate, and the words you care about are illegible in the corner. Filtering those out is what makes the visualisation informative.

The second layer is domain filtering. Your own brand name in a set of reviews about your product is technically the most frequent word and completely uninformative — of course it is there. Excluding it is not distorting the data; it is removing a constant.

The Remove button next to each word in the frequency table makes this iterative. Generate, spot the uninformative words, remove them, and the cloud gets more useful with each pass. Our Stop Word Counter shows how much of a text is made up of filler words in the first place.

How the sizing works

Font size is scaled by the square root of frequency rather than linearly. With linear scaling, a word appearing fifty times in a document where the next most common appears ten times renders enormous and everything else becomes uniformly tiny.

Square-root scaling compresses that range, so the most frequent word is still clearly the largest but the rest of the distribution remains readable. It is a deliberate trade: some precision in the size comparison, in exchange for a picture where more than one word is legible.

Words are placed largest first along a spiral running outward from the centre, and each position is rejected if it would overlap something already placed. The spiral is squashed to match the canvas shape — a circular spiral cannot reach the corners of a wide canvas, which drops words that would otherwise fit.

Reading a cloud without fooling yourself

• Check the frequency table before drawing a conclusion. If the top word appears twelve times and the next appears eleven, the size difference is not meaningful.
• Remember that longer words look bigger at the same font size. Compare the numbers, not the visual weight.
• A word cloud of a very short text is mostly noise — frequency needs volume to mean anything.
• Negation is invisible. If sentiment matters, the cloud tells you the topics and you still need to read the text.
• Dropped words are reported below the cloud. If a lot were dropped, reduce the maximum font size or word count rather than assuming they were unimportant.

Practical uses

Customer feedback is the most common one. A few hundred reviews or survey responses are too many to read carefully and too few to justify proper text analysis; a word cloud narrows it down to which themes deserve the careful reading.

Content auditing is another. Running your own article through it shows whether the words you intended to be central actually are. If your target phrase is barely visible, the article may be about something adjacent to what you think.

For more precise content work, our Keyword Density Checker gives exact percentages for a specific phrase, the N-gram Analyzer finds recurring multi-word phrases that a single-word cloud cannot show, and the Keyword Frequency Analyzer produces the full sortable table.

Frequently asked questions

Related tools