Best Practice Guide
The ultimate step-by-step guide for analytic content creators.
Usability best practices
Visualizations are intended to communicate data. As such it is very important that they are usable and meet their intended purpose. The usability of a chart is a key component of clear communication. The size, clutter and font size all affect readability and usability.
Charts need to be readable so not too small and given that size conveys importance not too big to overwhelm other data on the page – there is a balance.
Shrinking a chart with many components can cause the scale or individual components to be lost. If small charts need to be used consider using sparklines – purposefully small charts where the compromise on size is well understood.
<C34 image – do and dont>
When designing a dashboard consider the size of the chart relative to other objects on the screen. If the chart dominates the dashboard users will feel that that metric is more important than others on the page. Scale your charts to convey importance.
<C35 image – do and dont>
Charts should have minimal decorative elements applied to them. Each additional layer of a chart distracts the user from focussing on what is important. Most charts and visualizations use bars, lines, or data points to display values. These data objects should dominate the users attention. For other elements of a chart, such as titles, axis labels and background elements whilst they may provide some additional information they are not core to the data. As such they should be de-emphasized in your design.
Use Light colors for non-data Components
Non-data components of charts should be visually pushed to the background so that they do not distract the user from the data. The colors used should be light and the weight or thickness should be light as well.
Element color use Axis Lines Use thin gray lines of medium intensity Borders Whether around the graph as a whole, its plot area, or a legend, when borders are needed (and usually they are not), use thin gray lines of medium intensity. Background Use white Axis Titles Legend
Use gridlines to sparingly
Gridlines are often overused or over-emphasized in visualizations. However, there are times they are useful to use. Use gridlines when:
Measure small differences
You want to enable a user to measure small differences when data points are not ranked. Gridlines help to guide the across the axis and enable a more accurate comparison.
<C36 insert image – column chart by date – no lines, column chart lines – make sure columns are uneven and do not look like a trend or a ranked>
For charts that have a large number of data points, a gridline helps the user traverse the chart and interpret the data correctly.
<C37 insert image – show time-series line – no gridlines, time-series grid line – make sure 12 + months of data>
Enable users to focus on an area
This is generally used for scatter plots where users are comparing two numeric axis. Gridlines help to associate values on both axis simultaneously.
<C38 insert image – scatter plot no grid, scatter plot grid>
When using gridlines only use them for numeric axis and use a light color so as to not distract the user from the data.
<C39 Insert image – bad heavy gridlines x and y, light gridlines numeric>
Minimise Use of Labels
Labels are rarely needed in a visualization. The charts axis performers the function. Use labels only in cases where no axis is displayed such as pie charts.
<C40 insert image – bar chart labels, bar chart none>
Avoid Background Images
Background images clutter a chart and make it harder for a user to focus on the numbers.
<C41 insert bad background image versus none>
The exception to this rule is when the background image adds meaning – this could be in the form of a map for a scatter plot or an image of a heat map of a vehicle highlighting where damage occurs.
<C42 insert image map and car>
Avoid 3D Charts
3D charts do not add any additional analytic insights. They are heavy and draw the users attention away from what is important and can even obscure the results of a visualization.
<C43 IMAGE 3d bar versus bar>
Avoid using 3D charts, unless they serve a visual purpose and are interactive such as a body heat map which a user could rotate to see injury sites.
<insert video – human body image>
Avoid Drop Shadows
Drop shadows for the chart components or plot should not be used. They do distract and draw attention away from the data.
<C44 insert image drop shadow on bars, no shadow>
Drop shadows can be used very lightly for the entire visualization when being added to a dashboard. This can help to make the visualization stand out from the page.
< C45 insert image chart on dashboard with drop shadow>
Avoid Rounded Edges
Rounded edges on bar and column charts are best avoided. Apart from adding no additional value in terms of analysis, they can obscure results by making the actual values harder to interpret.
<C47 image rounded versus hard>
Similar to 3D effects chart animation has been seen as adding value to the analytical experience. It does not. Avoid animations and focus instead on charts load speeds. End users do not want to have to wait additional milliseconds watching their chart animate on load – they want to understand what the data tells them.
<C48 Insert Image – Animated gif versus Instant Load>
Don’t overload your charts
Keeping a chart as simple as possible and paring it down to its minimal needs is crucial to improving visual communication. This includes the number of series or values that are presented in a chart.
Reduce Categorical Items
Reduce the number of values presented in a bar chart. Either filter for top 10 (n) or bottom 10 (n) if ranking is important. This is useful when your users want to understand the best or worst-ranked items.
<C49 Insert bar chart lots, Bar Chart few>
Benchmark to average
If benchmarking a value to total is important then compare the value to the average. For example, if you want to understand how a single store compares to all other stores create a visualization that compares it to the Average or Top 5 Stores and Bottom 5 Stores.
<C50 Insert bar chart lots of values, versus compare to Average>
Use multiple series sparingly
The use of multiple series (metrics) in a single chart should be used sparingly. Only use when you are doing comparative analysis or looking for correlations. At most two series can be combined. Any more and the chart becomes cluttered and difficult to comprehend, even with dual axis.
<C51 Insert image multiple series versus 2>
Reduce color in Line, Stacked Bars and Pies
Displaying too many color series in a time series analysis or stacked bars makes analysis difficult. With stacked bars and pies 3 colors should be the maximum. With line charts up to 5 is OK.
Consider using two or more visualizations instead of too many color series.
<C52 Insert image bad and good line charts and stacked bars>
Font Size & Hierarchy
Font size is important to ensure that users can quickly understand the purpose and easily analyze the data in a chart. All fonts need to be of a size that are readable. Consider giving more weight to the chart title (this tells the user what the visualization is for) versus the axis labels and titles which must be readable but do not need to be overly prominent.
<Insert something about font hierarchy for each element in a chart – a simple table>
Consistent Chart Formats
Make sure that any formatting applied to a chart type is consistent throughout your dashboards and their subtabs.
Consistency in design ensures the user is not distracted by random changes in the way charts are presented.
<C53 insert image – a dashboard with lots of different styles, a tab with a single style>