What are the best ways to use data visualization to communicate?

It is important to insure visuals are usable, understandable, and bring value to our reports. Just because we can include a visual doesn't always mean we should. Do not take up space or overwhelm users with unnecessary visuals.

Choosing the right type of chart is a laudable task in data visualization as it can help the stakeholders and audience identify important trends and insights at a glance for critical business decision making. If you want to show geographical locations, maps will be exciting descriptors.

The purpose of Analytics is to tell the Audience something that may/will cause them to take action. It is too easy to overload or try to be too clever with colours and cut outs and 3d graphs. Focus on the need and purpose Enable analysts, insight and action. Keep it Simple

Keep it simple, use data labels and good annotation summary with any type of chart. With Power BI new features like custom labels, it became much easier.

Choosing the right way to represent a dataset depends on data type. Qualitative data works well with histogram, pie chart and donut charts. Quantitative data can be analyzed in a method that makes visualization clearer. Principal component analysis is useful to pick out the key variables that show the majority of variance in the data. In that case a simple Oxford plot could summarize a data set clearly along a x y axis.

The label takes a pivotal role in Data Visualization because it interprets a lot to the target audience/stakeholders. Using color correctly and logically contributes immensely to insights drawn from visualization.

One should also consider the mediums in which the visualization will be displayed when choosing the appropriate colors/labels, and symbols. For example, if a figure containing data points is the visual being used such as a scatter plot graph, and it is being presented in a power point, then it may be acceptable to display different data points using the same symbol such as a circle but infilled with different colour to represent the different sets (explained in legend). However, if the same data is also to be displayed/distributed in a grayscale/b&w/non colour medium such as a colorless print out, then the different sets of data within the visual must have a range of symbols representing them rather than colour infills.

The same could be said for legends and markers as well as line styles in a plot. When plotting time series data, e.g., take care to use appropriate short legends and use different markers and line styles to separate different aspects of your testing results to clarify, not overburden the reader of your work.

It helps to follow your organization's colour palette when choosing colours for your chart. I use the following techniques: 1. Bright colours (Eg - yellow, green, etc.) to highlight key data points. Try to use these colours on those areas where you want the reader's attention. Avoid using bright colours on the entire chart as it becomes jarring to the eye. 2. Secondary colours (eg - gray, dark blue, etc) to be used on those areas where you don't want your reader's attention 3. Stick to 2 colours per chart/slide. Too many colours causes a lot of distraction 4. Be consistent in your font size and colour 5. ALWAYS ALWAYS, showcase your insights from the charts. Add the key message either as chart title or summary below your chart

Always consider the WCAG web accessibility guidelines for color and contrast; it's not just an arbitrary "oh it should pop" it's about making sure your experiences can be seen and understood by all.

Less is always more, decision makers are not impressed by huge amounts of charts and visuals, they need to know what is important to them to inform their operational and strategic decisions. Focus on answering key questions, and keep engaging with the stakeholders to ensure you always provide the relevant answers.

Better to focus on one question/insight at a time. Good to use simple visuals but keep them ordered correctly to make it easy to understand the story/actionable insight. Multiple but simple visuals will lead more understanding and curiosity for the topic so will trigger more deep dive questions from stakeholders.

As with everything, simplify! I have seen so many very bright leaders sprinting down a fruitless path because the data they were looking at was presented in a complicated manner, using unfamiliar terminology, etc. If your data visualizations are so complicated that the viewer COULD draw many different conclusions, it’s much less likely that the visualization is going to achieve what you’re hoping it will. Start with the story and the edits needed to simplify will be clear.

Fantastic point! The audience shouldn't have to dig through the data to find the answer they are looking for - it should be right in front of them! My clients appreciate easily digestible information that they can understand at a glance and painlessly navigate in presentations to their stakeholders.

When curating dashboards it is always worth taking the time to work with end users and understand their expectations and define what best practice looks like especially if there are multiple users of the same type of role. For example, I work in manufacturing where we have over 100 team leads. Their role is the same across the business all be it they work in different areas and teams. Their role and expectations are the same so it makes sense to provide a consistent dataset regardless of where they are. Where KPIs are used, it's usually a good idea to have somewhere that defines what they are and how they're calculated whether it be a tooltip as you hover over, or an additional sheet as part of the visualizations

The best presenters are storytellers. There are 3 essential frameworks for crafting your story: 1) Aristotle's 3 Acts (setting, conflict, resolution); 2) SCQA (situation, conflict, question, answer); or, 3) the hero's journey. Start with your story and then spread it across your slides. One main point per slide in a logical flow. This is called "horizontal logic". Then examine each slide and add supporting evidence and visuals to support the main idea. This is called "vertical logic". Use action titles that clearly state your main idea on each slide, and limit written words to not more than 50.

All research tells a story, but not all researchers are great story tellers. The objective of a research report is to share the outcomes and recommendations in a manner were every reader can understand what was done, why it was done, and then replicate that project with ease if they choose too. Data visualization such as charts, graphs, and the headings of sections should be used to simplify the data that is being shared. These elements should visually take the ordinary person on a journey; from the initial problem, to the final outcome- allowing them to walk away with a clearer understanding.

There are 2 different concepts that I focus on when trying to tell a story with a reoccurring visual report, depending on the data. 1. Start with the big picture and get more granular. Start with a ‘Summary’ page that shows the main metrics, trends, and main attributes, then use 2-4 other pages for specific purposes to dig deeper into the data. Often times each pages gets more granular. 2. Think chronologically. What is the first question my end user will have, then when that gets answered and it leads to another question, how do I answer that? Both setups usually have a similar output where reports start more high level and keep letting the user dig deeper. Whichever concept works better for you, you should run with though!

In my experience, the story is the most important part of a data visualization. If you know the story that you want to tell, and you know the story that your data supports, you’ll be well on your way to effective visualizations. Finally, if you know and understand your audience’s pains, it’s much easier to use your narrative and data visualizations to show them where they can have gains.

Before you start creating the charts and the deck, create a story line. Either use a blank sheet of paper and write down the key points you would like to highlight in your deck. Or use a blank presentation and start adding titles to each blank slide. These titles will form part of your story line. Having a story line will help you to create the right set of charts and messaging and will keep your efforts focused.

One of the strong methods to test and refine data visualization is to formulate "zero team". The "zero team" should comprise diverse colleagues who have not seen that data or visualization before. Ask them Qs on what are they able to interpret from that data visualization. Based on the responses, assess the deviation from what they are understanding Vs what you intend to communicate through the data. Often we get too much into how to beautify the data visualization - this sometimes dilutes the "why" of the existence of the visualization. Hence, formulate the teams and keep refining the visualization till the time their understanding overlaps with your intent of the data visualization.

A helpful strategy for gauging the effectiveness of data visualizations is the 5-second rule. You should be able to draw high-level insights within 5 seconds of looking at the dashboard. If you can't, consider dialing back and refining it.

Along with the commonly understood notions of 'literacy' and 'numeracy', there is 'graphicacy', which similar to the other two is the ability to ingest and produce meaning via some content type. The skill level distribution for adults for text and numbers is somewhat abysmal, it is minuscule for graphic content - even for simple types like Excel spreadsheet chart types.