- The five best data visualization techniques to always remember when visualizing data include knowing your audience, focusing on your purpose, choosing charts based on what you want to show, specifically picking the right colors and aesthetics, and testing your work.
- More best practice tips on creating effective data visuals can be found here.
There are a number of techniques you can use for data visualization, but it can easily be overwhelming to pick the best for your upcoming presentation.
In a just few short minutes of online research you will quickly find that there are many strong opinions that surround a number of common techniques — all of which will make you unsure of exactly what technique is the most effective.
Here are the top five best practices to keep in mind when visualizing organizational data in an easy-to-understand way.
1. When deciding on what techniques to leverage in a data visualization presentation, you must first know your audience.
Information design isn’t about the science of data, it’s about people. You share visualizations in order to give useful information to others. So, it’s important to understand their needs, skills, and interests. Is your audience data literate? Will they readily understand charts you are showing them?
Some chart types may need some explaining — for example, I still have to think twice when looking at a bullet chart. The more you know about your users and their understanding of data, the better. You’ll be able to craft visualizations that work for them simply.
2. Shift your focus on the visual’s bigger purpose for it to be truly effective in decision-making.
When planning your visualization, have its purpose clearly in your mind. What is the visualization for? At this stage, don’t worry about the information you will show, but ask yourself: what decision will be made using this visualization?
After all, if the visualization will not help with some decision, why are you building it? Hopefully not just for decoration.
A good visualization provokes conversations and discussion about the topic; it doesn’t just answer questions, it raises questions too.
3. To know what chart type to use, ask ‘What do I want to show?’
Once you know your audience and have a good idea of the purpose of your visualization, it’s time to get down to choosing your chart types. What do you want to show?
To show how two or more values are related (for example, age and educational level) consider using a scatter plot to show both in a simple way.
Each data can be shown as a point or bubble positioned as an x,y coordinate representing the two values. Some scatter plots also allow you to change the size of the bubbles to show another value, such as income.
Change Over Time
If you are showing change over many time periods — more than just a few data points — then a line graphic is most useful. Also lines make it easy to plot multiple series together.
However, if you are showing change over just a few values — say changes over the last six months — consider using a bar chart again, unless you have multiple series.
This means showing how a single value (sales, perhaps) is made up of other values (say, for each region). The pie chart is a notorious example of a composition visualization. It has a bad reputation because so many pie charts are badly used.
It probably is best to avoid them unless you have a good reason. Certainly they are easy to understand, so long as they only have a very few values, and so long as you don’t want people to make exact comparisons at a glance.
If you want readers to compare values, bar charts of various kinds are very useful. You can easily compare between items.
4. The best data visuals use specific colors and aesthetic techniques to convey information, not just because.
Choose colors carefully to convey information. It’s a good practice to use one color for consistency throughout your document, web page or presentation. Introduce new colors when you want to make comparisons or show contrasts in a single visualization.
You can use shades of the same color to convey information. Shaded maps are popular, but shaded charts can be very effective too. Note that it can be difficult to make exact readings of a shade, so don’t rely on shading for precise information.
Multiple colors, as I said, can show contracts. But too many colors is confusing and almost literally sore on the eyes. If you are thinking of using more than 10 colors, think carefully. If you plan on using 20 — think again!
5. In order for your key stakeholders to get the information they need, it’s vital to test if your work gathers and displays this information accurately. And then test it again.
The most important best practice I can recommend is also the most overlooked. Test your work!
Visualizing data is a skill that takes time to get really good at. The last thing you want is for your work to be debunked due to ignoring a simple QA process.
It’s a really good idea to sit down with a typical user of your work. See what they like and don’t like. Ask them to explain to you what conclusions they can draw from your visualization. What questions does it raise? What questions does it answer? What conversations do you have when discussing the chart?
Perhaps make two or three versions of your visualization to try out different techniques. See what works best in practice by asking users.
As you get confident with your design, you’ll increasingly know what works well, but it’s good not to become complacent. Remember that visualization are not just about information — like all work, our focus needs to be on the people and their needs.
Donald Farmer is the Principal at TreeHive Strategy. He is an internationally respected thinker in the fields of data analysis and innovation, with over 30 years of deeply practical experience. His background is very diverse, having applied data analysis techniques in scenarios ranging from fish-farming to archaeology to advanced manufacturing. He has worked in award-winning startups in the UK and Iceland, and spent 15 years at Microsoft and Qlik leading teams designing and developing new enterprise capabilities in data integration, data mining, self-service analytics, and visualization.