How to structure data for different chart types

Start with dimensions and metrics

Dimensions

Dimensions describe the data and create categories.

Examples:

• City

• Year

• Segment

• Category

• Property type

These values are usually text labels.

Metrics

Metrics measure the data and provide the values shown in the chart.

Examples:

• Revenue

• Transactions

• Average price

• Min price

• Max price

These values must be numeric cells.

Bar chart

A bar chart needs:

• 1 required dimension

• an optional breakdown dimension

• at least 1 metric

Simple bar chart example

In this example:

• City is the dimension

• Listings is the metric

Stacked bar chart example

In this example:

• City is the main dimension

• Property type is the breakdown dimension

• Listings is the metric

Use a breakdown dimension only when you want to split values into subgroups.

Composed chart

A composed chart requires:

• 1 dimension

• exactly 2 metrics

• no breakdown dimension

This setup is useful when you want to compare two related numeric values in one chart.

Important: Do not add a breakdown column here. The structure must stay simple.

Line chart

There are two valid ways to prepare data for a line chart.

Model A: multiple metrics

Use this model when each data series can be stored in its own metric column.

This works well when the series are fixed and easy to compare.

Model B: breakdown structure

Use this model when you want to work with a breakdown dimension or support filtering later.

This structure is often more flexible for editorial use because it keeps categories in a dedicated column.

Which line chart model should editors use?

Use:

• Model A when the series are fixed and unlikely to change

• Model B when the categories may be filtered or reused later

In many editorial cases, Model B is the safer long-term choice.

Pie chart

A pie chart requires:

• 1 dimension

• exactly 1 metric

This chart works best when the number of categories is low.

Best practice: Keep the number of slices small. Too many categories make the chart hard to read.

Floating bar chart

A floating bar chart requires:

• 1 dimension

• exactly 2 metrics in separate columns

These two metrics represent:

• lower bound

• upper bound

Example

Important: The backend does not check whether the minimum is actually smaller than the maximum. Editors need to verify this manually.

Table chart

A table chart behaves differently from other chart types.

How it works

• no aggregation

• row order is preserved

Example

This makes a table chart useful when you want to display the data exactly as it appears in the sheet.

How to choose the right structure

Use this rule of thumb:

• one category + one metric = simple chart

• one category + breakdown + one metric = grouped or stacked chart

• one category + two metrics = composed chart

• one category + one metric with share of whole = pie chart

• one category + min and max values = floating bar chart

• exact row display without aggregation = table chart

Common structure mistakes to avoid

Adding extra helper columns without a purpose

Only include columns that are actually needed for the selected chart.

Using text instead of numeric metrics

A value like "7200" stored as text will not behave like a real metric.

Mixing chart structures in one sheet

Do not combine several logic models in the same dataset unless the chart setup explicitly supports it.

Using too many categories in a pie chart

Pie charts should stay simple and easy to scan.

Quick checklist

Before you connect the sheet to a chart, check:

• the dimension column is clear

• the metric columns are numeric

• the chart type matches the column structure

• no unnecessary columns are included

• the data model is consistent from top to bottom