Undoubtedly you’ve heard this advice when it comes to your data analysis and visualizations…
“…be sure to provide context.”
But what does this mean?
How do you even know what context to include? Or even when to include it?
Understanding where and when to add context is much easier when you have some ideas of the types of context that can be helpful when communicating with data.
These three common examples can be particularly useful...
1. Benchmarking
Imagine you’re presented with Key Performance Indicators (KPIs) from an entirely new department or even industry. What information would help you identify if these things are on-track?
We average 10 calls per day.
Our sales were $15,000 last month.
The error rate is at 2%.
Benchmarking helps answer the question “What does good look like?”
It can be as simple as a comparison to the average or something more sophisticated like a goal that has been set with growth or improvement in mind.
Examples of benchmarking:
A target set by your organizationEx. dollars in sales, # of calls, days per process step etc.
A comparison to a previous time period Ex. Average for that KPI over timeEx. Month-over-Month (MoM), Year-Over-Year (YoY)
A comparison to other high performing organizations within your industry
Including these benchmarks in your analysis and visualizations can help the viewer understand where-we-are-now versus where-are-we-headed or where-are-we-striving-to-be.
And while it can be affirming to demonstrate that a KPI is on track, it can be infinitely more valuable to identify an area where additional support might be needed to achieve the desired goals. A compelling example of how to drive outcomes and results through analysis!
2. Special Cause Variation / Planned Interventions
Now imagine you’re presented with a line chart that seems to be gently oscillating up and down over time. Then all of sudden there’s a dramatic change in the pattern. Maybe there’s a huge spike on one reading, or maybe the overall trend completely changes and starts to skyrocket or plummet.
The first question likely to come to your mind is, “What happened here?”
How helpful would it be if additional information was provided about something unexpected or out-of-the-ordinary that occurred around the same time?
Maybe there was a power outage that occurred on the same day the plummet occurred. Or maybe there was a sale just before things began to skyrocket. These are examples for potential sources of special cause variation.
Special cause variation refers to an unexpected variation in an established process that is different from any normal fluctuations that may be seen.
Including information about a potential special cause can help position the viewer to quickly make sense of what may have happened and better evaluate the appropriate next steps.
And even in cases where the data doesn’t demonstrate a change, it could be important to include information about planned interventions.
For example, imagine your organization has invested significantly to improve a key outcome. It might be very important to include this information on all monitoring going forward to help assess the effectiveness of that change.
Special Cause VariationEx. Root cause analysis - an unexpected outage, an unexpected windfall
Planned intervention Ex. Process improvement - updated process procedure, new marketing campaign, adding new resources
3. Common Scale
Finally, imagine that you’re presented with a bar chart that shows how many books each person has over the summer.
Would you immediately be impressed to see one person had read double the number of books compared to the next highest?
How would your impression change if you learned that their books had half as many pages?
Normalization can include the task of bringing data into a common scale without changing the shape of the data.
So in the example above, you might realize that even though one person has read double the amount of books, the page count is roughly the same. This can be very helpful to prevent the viewer from inadvertently overemphasizing a result.
Now, you may not run into a Pages-per-Book scenario in your analysis, but here are some common variables that you might want to be alert to as candidates for normalization:
Per Person, Per Capita: consider when comparing regions that vary greatly in population
Per Day: consider when comparing per months of different length (eg. January to February)
Per Unit: consider when comparing amongst groups of different resources
In Summary...
Adding context to your analysis and visualizations can help the viewer understand the data story quicker and with less chance of error. Think of it a way to help make sure your take-aways are as clear as possible!
Now that you know three common examples of context, consider if any of these will help your overall communication the next time you’re conducting data analysis or preparing a visualization.
Be passionate. Seek mastery. Learn with humility.
-Stacy
BLACK FRIDAY CAME EARLY!
Save up to 50% on Maven Pro plans today!
This week, we're offering up major discounts on individual subscriptions at Maven Analytics. Don't wait -- this offer ends soon!
Stacy Giroux
Cohort Learning Lead
Stacy is a former Cohort Learning Lead for Maven Analytics, helping to design, manage, and faciliate immersive bootcamp experiences for aspiring data professionals.