Visualise your data with truly private AI

Compare a value across categories; bars can be grouped or stacked by a second category.

Use it for

Ranking products, regions or teams by sales, counts or averages; group or stack the bars to add a second category.

Avoid it when

There are dozens of categories (show the top ones and group the rest), or the axis is time: a line chart shows change better.

Columns

Categories (category, yes/no or date)

A lighter bar chart for long rankings: one dot per category.

Use it for

The same comparisons as a bar chart when there are many categories and so many solid bars would feel heavy.

Avoid it when

Values are close together and readers need exact differences: labelled bars read more precisely.

Columns

Categories (category or yes/no)

Compare two measures for each category and show the gap between them.

Use it for

Before and after, or plan versus actual, for each category: this year's and last year's sales by region.

Avoid it when

You only have one measure, or more than two values per category.

Columns

Categories (category or yes/no) · First measure (number) · Second measure (number)

Show how positive and negative contributions add up to a total.

Use it for

Explaining how a starting value becomes a final one: monthly cash flow, a profit bridge, changes to a budget.

Avoid it when

The steps are not changes of one quantity that add up.

Columns

Steps (category, yes/no or date) · Change (number)

Show how a quantity shrinks through the stages of a process.

Use it for

Stages that narrow down in order: visits, sign-ups, trials and purchases.

Avoid it when

The stages do not follow a sequence or do not shrink: use a bar chart.

Columns

Stages (category or yes/no)

Show how a value changes over time or along an ordered axis.

Use it for

Monthly revenue, daily temperatures, weekly sign-ups; one line per series to compare trends.

Avoid it when

The axis is unordered categories (use bars), or there are more than about seven lines, which tangle.

Columns

X axis (date, number, category or yes/no) · Value (number)

Show change over time as filled areas; stack the series to show how they add up.

Use it for

Totals over time and the series that make them up: sales by channel, month by month.

Avoid it when

The series cross a lot, or each one must be read precisely: lines are clearer.

Columns

X axis (date, number, category or yes/no) · Value (number)

Show tasks or events as bars between a start date and an end date.

Use it for

Project plans, schedules and anything with a start date and an end date per task (a Gantt chart).

Avoid it when

Your data has a single date per row rather than a start and an end.

Columns

Task (category or yes/no) · Start (date) · End (date)

The sample has no start and end dates: try it with your own file.

Show the open, high, low and close of a price for each date.

Use it for

Prices with an open, high, low and close for each period: shares, cryptocurrencies, commodities.

Avoid it when

You only have one price per date: a line chart is enough.

Columns

Date (date) · Open (number) · High (number) · Low (number) · Close (number)

Show the share of a whole held by each of a few categories.

Use it for

A few parts of the same total: the market share of five brands, how a budget is split.

Avoid it when

There are more than six or so slices, or they are of similar size: bars compare them better.

Columns

Slices (category or yes/no)

Show parts of a whole as nested rectangles, with an optional second level.

Use it for

Hierarchical totals with many parts: sales by category and subcategory, disk usage.

Avoid it when

Readers need to compare exact sizes: areas of rectangles are hard to judge.

Columns

Top level (category or yes/no)

Show a hierarchy of parts of a whole as rings.

Use it for

Hierarchies read from the centre outwards: region, country, city.

Avoid it when

The hierarchy is very deep or very wide: the outer rings become too thin.

Columns

Inner ring (category or yes/no)

Show a hierarchy of parts of a whole as stacked layers.

Use it for

The same hierarchies as a sunburst, in bands that are easier to label.

Avoid it when

There is no hierarchy: a single level reads better as bars.

Columns

Top level (category or yes/no)

Show how the values of one measure (or of dates) are spread.

Use it for

The shape of one measure: customers' ages, order values, response times.

Avoid it when

There are very few rows (fewer than about twenty), or the variable is a category: count it with bars.

Columns

Measure (number or date)

Compare the median, spread and outliers of a measure across groups.

Use it for

Comparing medians, spread and outliers across groups: delivery times by carrier.

Avoid it when

Your audience is not used to box plots, or the groups have very few values.

Columns

Measure (number)

Compare the full shape of a measure's distribution across groups.

Use it for

Comparing the full shape of distributions across groups, for example when one group has two peaks.

Avoid it when

The groups are small: the smoothed shape suggests patterns that are not there.

Columns

Measure (number)

Show every individual value of a measure, group by group.

Use it for

Seeing every individual value in each group, when there are not too many points.

Avoid it when

There are thousands of points per group and they pile up: use a box or violin plot.

Columns

Measure (number)

Show what share of the values falls below each level of a measure (ECDF).

Use it for

Reading percentiles directly: what share of orders arrives within three days?

Avoid it when

Your audience expects a histogram: cumulative curves need a sentence of explanation.

Columns

Measure (number)

Show the relationship between two measures, one dot per row.

Use it for

Two measures per row: price against rating, ad spend against sales; spot correlations and outliers.

Avoid it when

One of the variables is a category: use a box plot or a strip plot.

Columns

Horizontal axis (number) · Vertical axis (number)

A scatter plot in which the size of each bubble shows a third measure.

Use it for

A scatter plot with a third measure as size: countries by GDP, life expectancy and population.

Avoid it when

The sizes are similar, or the bubbles overlap heavily.

Columns

Horizontal axis (number) · Vertical axis (number) · Bubble size (number)

Show where points concentrate when there are too many to plot one by one.

Use it for

Seeing where points concentrate when a scatter plot has too many to tell apart.

Avoid it when

There is little data: plot the points themselves.

Columns

Horizontal axis (number) · Vertical axis (number)

Show which numeric columns move together, as a colour-coded grid.

Use it for

Seeing at a glance which pairs among many measures are correlated.

Avoid it when

Correlation is not causation, and it misses non-linear relationships: check promising pairs with a scatter plot.

Columns

Measures (2 or more: number)

Plot every pair of measures against each other at once.

Use it for

Exploring every pair of a handful of measures at once.

Avoid it when

There are more than five or six measures: the panels get too small.

Columns

Measures (2 or more: number)

Compare many measures at once, one line per row.

Use it for

Comparing rows across many measures: product profiles, patterns of high and low values.

Avoid it when

There are so many rows that the lines turn into noise: filter, or colour by a category.

Columns

Measures (2 or more: number)

Show how rows flow between the categories of two or more stages.

Use it for

Flows between stages: traffic source, landing page and outcome; budget from department to project.

Avoid it when

Each stage has too many categories and the diagram tangles: group the small ones.

Columns

Stages, in order (2 or more: category or yes/no)

Show how combinations of categories relate across several columns.

Use it for

How several categorical variables combine: customers' plan, channel and region.

Avoid it when

The variables are numeric (use parallel coordinates) or have many categories.

Columns

Stages, in order (2 or more: category or yes/no)

Show a value for every combination of two categories (or a category and a date) as a colour grid.

Use it for

A value across two categories: sales by weekday and hour, satisfaction by team and quarter.

Avoid it when

One of the axes has hundreds of categories.

Columns

Columns (category, yes/no or date) · Rows (category or yes/no)

Show one headline number, optionally compared with the previous period.

Use it for

A headline figure for a report or a dashboard: total revenue, average rating.

Avoid it when

The figure means little without context: add a comparison or a trend chart.

Columns

Any file: it counts rows or sums a measure.

Compare the profile of a few groups across several measures.

Use it for

Comparing a few items across several measures on the same scale: a product's ratings on five criteria.

Avoid it when

The measures have very different scales, or there are many items: grouped bars read better.

Columns

Axes (3 or more: number)