These best practices reflect recommendations shared by a cross-functional team of seasoned Lookers. These insights come from years of experience working with Looker customers from implementation to long-term success. The practices are written to work for most users and situations, but as always use best judgment when implementing.

This article explains how to choose a visualization type that will be most easily understood by users, based on the kind of information displayed.

Visualizations are an effective way of communicating insights found in data. Choosing the right visualization is a vital component of effectively sharing information across an organization. Here are a few overarching things to consider:

• Think about how end users viewing a visualization will interpret and take action on what they see. Be sure to focus on actionable metrics over vanity metrics.
• Use the simplest visualization possible to communicate a message. Overcomplicated visualizations are difficult to understand quickly, and are easy to misinterpret.
• Label axes and measures clearly.
• Use consistent colors (use a built-in color palette or implement a custom color palette to implement consistent coloration).

Below are some common visualization types, along with a few best practices to consider for each category. For a complete list of visualization options, please check out Looker's Visualization Types documentation.

Column and Bar Charts

Column charts and bar charts are effective for comparing two or more concrete categorical values. Column and bar charts enable end users to quickly compare lengths and then grasp the relationship between the categories presented. Column and bar charts differ mainly in orientation: bar charts are oriented horizontally, while column charts are oriented vertically.

• Consider using a bar chart over a column chart when dimension axis labels are long.
• Consider choosing a bar chart over a column chart when displaying negative values
• Always choose a column chart over a bar chart when comparing values over time.
• Use horizontal labels whenever possible, to ensure they are readable.

The measure axis should typically start at zero to avoid misleading viewers, but can be unpinned from zero in circumstances where the insight being shown would otherwise be completely lost (such as very small differences or changes across data points).

Line and Area Charts

Line charts and area charts are best for displaying continuous data, such as time. Discrete data points will be plotted, but those points are then connected, representing continuity between them. Both line and area charts facilitate trend analysis. Although these charts are similar, they should not be used interchangeably. Line charts are best used for comparing performance among groups or for showing more than one measure. Area charts are best for showing cumulative, part-to-whole relationships.

• Be thoughtful when stacking line charts, as this is easy to misinterpret because there are no cognitive clues to indicate that the line values are cumulative.
• Be thoughtful when using area charts for comparing individual groups, because the colors in the back quickly become obscured by the overlapping colors in front.

It's best to start the y-axis at zero, to ensure no misinterpretation of the data. If you need to zoom in on a particular trend, you can always place both versions together on a dashboard to ensure that viewers have the full picture.

Stacked Charts

Stacked charts, such as stacked bar charts or stacked area charts, enable you to add complexity to a visualization, as each consecutive series will be rendered above the last. However, these can also be difficult to read when not implemented properly.

• Avoid using stacked charts to visualize measures that shouldn't accumulate, such as averages.
• Avoid plotting too many lines or categories at once (five or fewer is ideal).

Dual Axis Charts

Use dual axis charts to visualize the relationship between two different measures. This may help to show trends or correlations present between those measures, that otherwise would not be obvious if those measures were plotted separately. These are often used when combining measures on very different number scales, such as total values with percentages.

• Consider combining different visualization types (such as a line and a bar) to clearly illustrate each measure.
• Use contrasting colors for each measure, to further distinguish between them.

It may be useful to consider your users' language context when you decide where to put the primary measure. For example, people who are used to reading from left to right will tend to look to the left first, and so it would be useful for them to find the primary measure on the left-side y-axis (and vice versa for those who are used to reading from right to left).

Pie and Donut Charts

Pie charts and donut charts can be used to show how a value or measure is split out across categories.

Because segments in these charts should always add up to 100 percent, pie and donut charts provide viewers with a sense of the proportions of the pie that can be attributed to each category. These types of charts can be great for showing the general composition of the data, but they should not be used for comparing individual sections to each other or for representing exact values.

• Include fewer than five categories in the pie or donut chart whenever possible.
• Choose visualizations like column and bar charts over pie and donut charts whenever possible, as differences in angles and circular areas can be difficult for the brain to detect. Column charts and bar charts both have a stacked percentage plot option for representing series as percentages stacked on the y-axis, where all values add up to 100 percent.

For more on choosing the best visualizations for the data you want to display, please check out our eLearning course on Designing Great Dashboards. For detailed instructions on how to implement each visualization type, please see our Visualization Types documentation.