This is the third in a series of data visualization lessons I learned before I was a teenager. If you want to catch up, in sixth grade I learned to know how you’re being measured and in seventh grade I learned that if something seems off, it probably is.
In first grade, the biggest incentive in my life was earning gold stars next to my name on the class board. These elusive gold stars could be handed out for a variety of reasons, but usually they were earned for answering questions correctly in front of your peers. The way the five and six-year olds in the class tried to earn these gold stars, you would think they were worth a million dollars each.
One fateful day, the teacher introduced the story of The Tortoise and the Hare. After explaining what a tortoise and a hare were, she asked “For one million dollars… err… one gold star…. who do you think will win the race?”
This was my moment. Little did the teacher know that I already heard this story in kindergarten. I was not going to be fooled by the seemingly obvious choice that the faster rabbit would prevail. This was review and possibly the easiest gold star I would ever earn. I practically fell out of my chair trying to get the teacher to pick me.
That’s when I learned one of my best data visualization tips.
Despite knowing the correct answer, in my haste and impatience answering the question, I blurted out “The hare!” I realized my mistake immediately, but it was too late. Having answered the question incorrectly, I did not earn a gold star and the teacher moved on to the next student.
The lesson that I learned was: Be patient before taking action on an insight.
There are many times as an analyst when it is tempting to act too soon. Sometimes you think you’ve seen something similar before and assume the same phenomenon is taking place this time around. Sometimes we have a funny way of finding a narrative that fits, even when there isn’t much evidence to support it.
Unfortunately, if you act too soon and take the wrong action, you often have little to no time to reverse the damage. In my extreme case as a six-year old, the ‘train left the station’ after about two seconds and I had missed my opportunity. Even if you have the chance to change direction after making a wrong decision, you risk losing credibility with your stakeholders and may incur costs as a result of your inefficiency.
While data visualization has the ability to make insights emerge, sometimes the first layer doesn’t tell the full story. So what can we do?
First, be positive you’re starting with a strong foundation. With the Decision-Ready Dashboard framework I use in my consulting partnerships, there are two quality assurance steps; one after the data preparation phase and one after the dashboard development phase. This duplicative quality assurance ensures that any decisions that are eventually made are based on sound data.
Next, dig deeper. As is the case in my Super Sample Superstore dashboard, I often talk in terms of ‘descriptive’ and ‘prescriptive’ analytics. I estimate that 80% of recurring reports are descriptive in nature, providing only a ‘10,000 foot view’ of what happened. Before jumping to conclusions, you should dig deeper into a prescriptive layer to figure out why it happened. If the prescriptive views do not exist, you should still only use the descriptive reports as a starting point for an analysis. After you find a lead, do additional discovery or ad-hoc analytics to find context that supports the insight.
Lastly, collaborate. If you have the opportunity, bounce your insights off of your colleagues. Not only will this support system help determine if your insights make sense, but you might learn something new to make them even better. Maybe there was a marketing campaign you didn’t know about that helps explain the story in the data. Maybe a veteran analyst knows about a seasonal event that occurs around the same time every year.
By exercising patience and finding additional context that helps support your insight, you maximize the chance of the insight causing action.
Thanks for reading,