What are they?
Data visualisation is the graphic representation of data and information, and it is particularly efficient when dealing with large or complex datasets.
Data visualisation is often exploratory, as it encourages the user to familiarise themselves with the visualisation in a non-linear way and to look at the data from whichever angle they want.
Done well, data visualisations:
- Encourage data exploration
- Grab attention with visual metaphors
- Retain users’ attention with smart interactive features
- Allow users to filter, sort and manipulate data visually
Process
We can create data visualisations based on two types of data:
1.
QUANTITATIVE DATA OR NUMBERS
- E.g. quantity, frequency and timelines
2.
QUALITATIVE DATA
- E.g. process diagrams and flowcharts
Longitude examples:
Other examples:
Seven functions of data visualisation
When creating a data viz, think about what it’s for. Do we want our user to make comparisons across quantities, look at the overall distribution of data, compare parts with the whole or explore data via geographical coordinates?
Here are seven purposes of data viz and which formats to use for each:
- Comparison: bar chart, proportional area chart, packed circle, stacked area chart
- Show concept: venn diagram, flow chart, linear process, matrix diagram
- Distribution: treemap, alluvial diagram, sankey diagram, bubble scatterplot
- Parts-to-whole: doughnut chart, stacked bar chart, funnel chart, marimekko chart
- Correlation: heatmap, chord diagram, arc diagram, network visualisation
- Trend-over-time: line graph, stream graph, timeline, slope chart
- Geographical: choropleth map, flow map, bubble map, dot density map
Sources: datavizproject datavizcatalogue
How to create a data visualisation
1. Format your data sheet
Your data should be legible at a glance. The narrative should be simple to read, and all of its elements easy to recognise. Make sure the title of the spreadsheet is logical and includes any necessary project information.
Include all the data, and just the data required for the data visualisation. Everything that appears on the datasheet should be used by the designer in some way (title, headers, data values, annotations, legend, sources and disclaimers).
When you format your data, keep things simple. Try to avoid colouring the cells or text, and use a standard text style and size. Consider making header rows and columns bold to distinguish them from the rest of the data content. And make sure that all your labels have been sub-edited to avoid any typos or inconsistencies.
2. Simplify multiple datasets
If you have multiple datasets or complex datasets that require multiple views, consider using separate tabs. These allow you to present the chunks of the data that will be turned into standalone data visualisation modules.
And if you use multiple tabs, make sure that the tab titles (the labels at the bottom of Google Sheets or Excel) have a clear and consistent name — it’s best, for instance, to avoid long titles as the tabs will only show the first few words before truncating.
If you have a lot of tabs, consider adding a summary tab as a contents list so we can navigate your data easily (you can also include hyperlinks for even quicker navigation).
3. Make chart titles clear
It’s best practice to start your dataset with a title that is bold and slightly larger than the rest of the text. This title may be different from the overall dataset filename title — especially if you have multiple datasets within the same file.
If the data visualisation needs a subtitle, please include this below the title in a smaller text size.
For more on how to write good chart titles, see new ways to write reports.
4. Include header rows and columns
Datasets must have clear header rows or column titles because these will form the labels for filters, and they allow the datasheet to be navigated easily during development. Make sure these are obvious by using a bolder text weight, and consider ‘freezing’ the column or row for ease of navigation — this is especially useful for larger datasets.
5. Organise your data
When you put your data in rows and columns, display it as close as possible to the way you think the data should be visualised. For example, if the data includes a time series, it makes sense to lay the data out via columns rather than rows because we read time horizontally rather than vertically.
6. Avoid data gaps
When you showcase your data in a table, avoid gaps and empty cells. If a value is zero, make sure the cell contains 0.
If a value is missing, you need to flag this in the dataset — perhaps with an extra note for designer, or with annotations that need to be included in the chart.
7. Use the right units
When you are formatting cells with numbers, make sure you use the correct unit of measurement — currency, weight or number of hours, for example. Make sure you’re consistent across the datasheet, and use abbreviations and decimals consistently ($2m, £1bn, etc).
8. Annotate your extra information
If a chart needs a special annotation or callout (for example, a label in the middle of the chart that highlights a specific data point), please include this in the dataset in a dedicated column or row.
9. Footnote your disclaimers and sources
Include any sources and disclaimers as footers in the dataset, below the data. Try to avoid including asterisks in cells containing data, because this could cause problems when extracting data for plotting and prototyping.
1. Think of your audience
Your target audience should be your focus throughout the design process. The tone of voice and use of visual cues should be tailored for the type of audience your data viz will address.
Is the information on display simple (or advanced) enough to satisfy your viewer’s attention without leaving them confused or overwhelmed?
It’s good practice to test your data viz with a colleague who only has limited context on the subject, and evaluate how they process the information they see.
2. Balance function and elegance
Elegant and beautiful data viz can draw the attention of the user. But it’s important to balance the aesthetics with a visual solution that elevates the primary function of data viz.
Whereas the core function is to show electoral trends or rising temperatures, it’s important to clarify the primary goal of your data viz, and ensure it does one thing well instead of many things poorly.
3. Remember accessibility
Alongside function and elegance, a third requirement is accessibility. Never pursue the first two at the cost of accessibility.
Remember that some users xxxxxxxxxxxx, and others xxxxxx. This affects your choice of colours, the size of your text and the clarity of the legend.
And make only reasonable demands of users. Some visualisations need to be quick and simple, and others can be more sophisticated to make sense of a complex subject.
4. Use colour with purpose
Colour selection should always be strategic. Introducing colour can be useful to organise categories of data, while linear or divergent colour palettes can depict a specific type of dataset, such as global temperatures or electoral results on a map.
Try to use colour sparingly and only when necessary. And keep in mind that some people are colour blind, so make sure that the colours of your choice are distinct enough and work for these users.
5. Encourage exploration
Data viz lends itself to exploration — especially when rich datasets with compelling stories are laid out effectively, making it interesting (and fun) to dive deep into the data to spot trends, patterns and outliers.
As a designer, use tools at your disposal to make the exploration playful and engaging, and not daunting or difficult. You can use clever annotations, clear labels and interactivity to invite the users to get lost in the data — without feeling lost.
6. Add interactivity
Digital gives us a new way to access and engage with data viz. Filters, toggles and hoverable features make digital data viz a much richer experience — one that can feel immersive and multidimensional.
It can be tempting to add interactivity to a data viz even when it isn’t necessary. There are plenty of examples of static data visualisations that work extremely well without the need for interaction, so ask yourself if an interaction is really necessary or if you’re overcomplicating things.
How to be a good partner to the design team
Here’s how to prepare your data for design:
1.
Remove any comments and track changes from the script.
2.
Use square brackets for any text that shouldn’t be included in the document. You can use this text to tell the design team what is:
- The title
- The subtitle
- A standfirst
- A top-level/section heading and lower-level headings (if it’s complicated, you can use an A/B/C system to show the hierarchy of headings)
- A boxout/sidebar
- A call to action
3.
Format your footnotes consistently, or if you’re using hyperlinks place them on the right text.
4.
Suggest pullquotes and pullstats using comments in the Word file. It’s a good idea to format these correctly in your comment box — i.e. put the name and job title underneath the full quote.
5.
Provide a clean, proofread data sheet, which could be linked from the charts in the Word file. Make sure that the design team knows exactly where the data is for each chart.
6.
Charts should have only the key storyline data included. Make sure the data sheet matches this and the labels have been proofread.
7.
Designers should not be rekeying/typing any words, because that’s how errors creep in. So don’t just provide screenshots/images of your charts or tables — make sure they can just copy and paste the text from the data sheet.
Real thought leadership.
Head office FT Longitude
For brands with big ideas.
© Longitude Research Ltd | Privacy Policy | Cookie Policy | Tax Policy
Bracken House 1 Friday Street London EC4M 9BT
