PlotSet LogoPlotSet
All chart types

Word Cloud Generator

Turn a list of terms and their counts into a word cloud you can style, embed and export. Upload a spreadsheet, set the size range and rotation, and publish — free.

Create a Word Cloud — freeSee the gallery

What is a Word Cloud?

A word cloud lays out a set of terms so that each one's font size reflects a number attached to it — usually how often it appeared in some body of text, but it can be any weight at all: votes, mentions, hours logged, tickets filed. The layout algorithm then packs the terms together, rotating some of them, until they fill the space. What you get is a picture of which words dominate a corpus, readable in about two seconds and from across a room. What you do not get is a chart anyone can read a value off, which is worth being clear about before you build one.

When to use a Word Cloud

  • You want the gist of a body of text at a glance

    Survey free-text answers, support tickets, review comments, a set of job ads. A word cloud answers 'what is this pile of text mostly about' faster than any other format.

  • The audience is being introduced to the data, not analysing it

    Conference slides, a report opener, a social post, a workshop warm-up. Word clouds are good at getting people interested in a corpus they have not read.

  • There is one obvious winner and you want it felt, not measured

    When a single term is genuinely dominant, sizing it fifty times larger than the rest lands harder than a bar three times longer than the next.

  • The terms themselves are the point

    Sometimes the interesting finding is which words appear at all — the vocabulary, not the counts. A cloud shows a lot of distinct terms in a small space, which a bar chart cannot.

When not to use a Word Cloud

  • The reader needs to compare two terms

    Font size is a poor encoding: people judge it inconsistently, and it competes with word length, so a long mid-weight word can out-shout a short heavy one. If the comparison matters, use a bar chart.

  • You are showing change over time

    Two clouds side by side cannot be compared — the layout is different, the scale is different, and nothing sits in the same place twice. Use a bump chart or a line chart of the terms you care about.

  • The counts are close together

    If your top twenty terms are all within a factor of two, every word comes out roughly the same size and the cloud says nothing at all. That is a ranked bar chart or a table.

  • The words need context to mean anything

    'Not', 'never' and 'without' get stripped or floated free, and a cloud of single words cannot tell 'good service' from 'not good service'. For sentiment, count phrases, or read the text.

How to make a Word Cloud

From a spreadsheet to a published chart, step by step.

01

Count your words first

The template expects two columns — the word and its weight — so the counting happens before PlotSet sees the data. A spreadsheet pivot, a COUNTIF, or any text-analysis tool will produce it. This is the one step people are surprised by, and it is also where you get to make the decisions that matter.

02

Strip the words you do not want counted

'The', 'and', 'a' and their friends will otherwise be the biggest things on your chart. Remove them, then look again — every corpus has its own filler, and in support tickets 'issue' and 'problem' are as empty as 'the'.

03

Open the word cloud template

Sample data comes bound, so you can see the exact shape of spreadsheet it expects before you upload your own.

04

Upload or paste your two columns

Import a CSV or Excel file, paste straight from a spreadsheet, or connect a Google Sheet. Bind the text column to the word and the number column to the value.

05

Set the size range and the rotation

Minimum and maximum font size decide how loud the contrast is; the angle range and step decide how many words are turned on their side. Keeping rotation to 0 and 90 degrees, or switching it off entirely, is easier to read than free angles.

06

Publish, embed or export

Publish for a live embed, or export a PNG, SVG or PDF. Export as SVG if the cloud is going into print — the text stays sharp at any size.

Start from the Word Cloud template with sample data already loaded, then swap in your own.

Open the template

The parts of a Word Cloud

Term
One word or phrase. Nothing forces these to be single words: 'customer service' as one term is usually more informative than 'customer' and 'service' apart.
Weight
The number behind each term — the count, score or frequency that drives its size. It is the only quantitative thing in the chart.
Font size scale
The mapping from weight to type size, bounded by the minimum and maximum you set. Widen the range for drama, narrow it when the counts are close.
Rotation
The angles words are allowed to take. Rotated words fill space more efficiently and are measurably slower to read; the trade is yours to make.
Layout
The packing algorithm that places each term in the first free space it finds, largest first. Position carries no meaning — a word near the centre is not more important than one at the edge.
Colour
In this template a gradient between two colours. It is decoration unless you deliberately map it to something, and a legend is what turns it from decoration into data.
Stop words
The terms you excluded before counting. They are not visible in the finished chart, which is exactly why they belong in the caption.

Common Word Cloud mistakes

Leaving the stop words in

A cloud whose largest word is 'the' has told the reader nothing except that you did not clean the data. It is the most common word-cloud failure by a distance.

Treating size as if it were readable

Readers cannot recover counts from font size, and they systematically underestimate large ones. If a number matters, put it in a caption or a table next to the cloud.

Forgetting that long words look bigger

'Implementation' at weight 40 covers far more ink than 'cost' at weight 60. The encoding is height, but the impression is area, and the two disagree.

Splitting phrases into single words

Tokenising on spaces turns 'climate change' into two terms that individually mean less. Count the phrases you care about as units before you upload.

Colouring at random and calling it a variable

If colour is not mapped to anything, say so — or map it to a second measure such as sentiment or category and add a legend. A random palette that looks meaningful is worse than an obviously decorative one.

Using one where a bar chart was asked for

If the question is 'which term is most common and by how much', the honest answer is twenty ranked bars. Word clouds are for attention, not for analysis.

Word Cloud: common questions

Bring a two-column list — the word and its count — then open the word cloud template and upload or paste it. Bind the text column to the word and the number column to the value, set the font size range, and publish.

Make your Word Cloud now

Free to start with 300+ templates. No credit card required.

Create a Word Cloud — freeAll chart types