Be honest, how many times have you or someone else in your organization said the following:
- “We have all of this data, but nothing is actionable!”
- “We’re suffering from ‘analysis paralysis;’ we need to take action.”
- “We lack the data / tools / talent to allow us to take data-driven action.”
I can’t count the number of times I’ve heard someone lament about the disconnect between the data available to them and the actionable guidance they’re hoping to get. If only I had more data, better data, integrated data, better tools, more user-friendly tools, better talent, or talent who understands my business! If I only had access to those things, I would be able to take data-driven action!
<April raises an eyebrow and gently waves a B.S. flag…>
In over a decade in the field of data science, I have worked with companies of all levels of sophistication….and I’ve heard this complaint at nearly all of them. I’ve seen the same issue at start-ups and Fortune 500 companies, companies with and without integrated data warehouses, companies with and without data scientists on staff, and companies with and without BI and data visualization tools.
This problem is pervasive.
We all hear about these mystical organizations that are inherently “data-driven.” We hear about companies who make their decisions and take action based on data, and we assume they have access to more or better data / tools / talent.
The hard truth
What if I told you that it’s not the availability of data, tools, or talent that’s allowing these organizations to close the gap between data and action?
Instead, what if it’s simply that they’re following a systematic process for turning observations from their data into insights, and turning insights into action?
What if you could learn that process too? Are you with me, Neo?
This problem is so pervasive, and companies are missing such a huge opportunity to take data-driven action, that we Sprocketeers have put together a systematic approach that you can use to take action from your data.
Our approach involves asking a systematic set of questions as you comb through the reports and analyses that you have available to you and then using the insights that arise in a test-and-learn framework to take data-driven action.
Step 1: From Observation to Insight
Almost every company has access to some kind of business intelligence, data visualization, or standard reporting tool. Most CRM’s have their own baked-in reporting platform to help companies understand their customers. Google Analytics reporting helps companies analyze and understand their web traffic. Some companies have access to integrated reporting through tools like Tableau, Business Objects, Power BI, MicroStrategy, DOMO, or a growing number of other BI and visualization tools.
These tools allow you to make observations using data. Observations are simply statements about a pattern observed in the data. For example, when one monthly bar on a net revenue bar chart is taller than the other one, we observe that “our net revenue went up from last month to this month.”
In my experience, most companies do a great job of generating observations from their data. However, many companies stop there, never generating insights from their data observations, and unfortunately, observations themselves are not actionable. Insights move beyond observations to include some essence of causality. Many times, they include the word “because.” The void between observations and insights is a vast one, and it is where many companies fall short in the quest to take data-driven action.
Who? When? Where?
The next time you’re looking through a set of standard reports, first I want you to ask 3 questions: Who? When? Where?
- Who shows that pattern? Is it more/less pronounced for certain people? (i.e., Drill into segments of customers to look for more refined patterns.)
- When did that pattern occur? (i.e., Examine trends & discrete observations over different chunks and periods of time.)
- Where did the pattern occur? (i.e., Drill into different geographic regions, business units, or other location or business segments to look for more refined patterns.)
For extra credit, you can even combine 2 of the 3 questions to see if the pattern changes for a particular customer segment during a certain season of the year, for example.
“Our net revenue is up because of a spike in sales on the 17th of the month at the San Diego store in our retail chain.“
You can see how once you’ve asked and answered that set of questions, you’ll have a much more refined set of observations. However, you still won’t have an insight. Remember, an insight gets at the likely root cause of the observation, and to do that, you’ll likely have to start hypothesizing beyond what your data can explicitly tell you.
A very common reason why companies stop short of generating a true insight is that they expect the data to explicitly tell them everything. A wise man once said, “You can’t analyze your way into the future.” Said another way, if your data observations are dry land, at some point, you’re going to have to lose sight of the shore.
Many times, you’ll need to use all of the data, business, and industry knowledge you have to infer the most likely cause of the observation you’re seeing in order to generate your insight. Data is powerful, but it rarely contains all the answers. Here’s how to do it.
Why? Why? Why? Why? Why?
If you’ve ever hung out with a 5-year-old for any amount of time, you know that they’re masterful at asking the question, “Why?” Can it be annoying? Sure. But, man, those little buggers can get a lot of information out of a person with their persistence!
It’s time to let that inner 5-year-old out, and let them loose on your data. Why do you think you’re seeing that observation? From there, I want you keep asking “why” until you stumble upon a causal factor that is under your control. When you get there, you’ll have landed on your data insight – the “because” underneath the observation.
Here’s what I mean:
- Why do you think net revenue was up on the 17th in the San Diego store? Because that location held a customer appreciation event on the 17th.
- Why do you think the customer appreciation event drove additional sales? Because customers were incented to hit a 1-day purchase threshold at the customer appreciation event.
- Why do you think the threshold incentive caused customers to buy more? Because customers could use a special reward coupon only if they spent above a threshold.
- Why would the reward coupon cause customers to spend above the threshold? Because customers experienced loss aversion when they were given the coupon, so they spent above the threshold to not miss out on the 1-day reward.
We have now closed the gap between observation and insight. “We have good reason to believe that our revenue went up because a customer appreciation event at the San Diego store created a sense of loss aversion in our customers, leading them to spend more than they typically would on that given day.”
Step 2: From Insights to Action
Once you’ve generated insights from your data observations, you simply have to take one more step to get yourself to that coveted step of taking data-driven action. To move from insight to action, you simply ask one more question: How might we…
“How might we create a sense of loss aversion in our customers at other locations and in other times of the year to drive a similar increase in sales?”
The question “how might we” allows you to generate ideas and hypotheses about actions you could take to replicate the insight you derived and cause an outcome similar to the original observation.
To help you figure out what your next action should be, we recommend taking a test-and-learn approach toward generating hypotheses and small experiments to test the validity and impact of your insight to ensure that the same results will be seen on a greater scale, at a different period of time, and/or for different groups of people.
“We could test the validity of a company-wide loyalty program with expiring rewards that creates a periodic sense of loss aversion in our customers, driving sales throughout the year across all locations.”
See how it all fits together?
It’s Your Turn
It’s time to get to work. Look around. You probably have some kind of report or chart within reach right now. Get going and start turning your data into action!