On Finding Actionable Insights

Despite having thousands of available resources that teach us how to manipulate data and create visualizations, not many teach the actual process of finding data that can translate into real business value. Revealing actionable insights from a dataset is incredibly challenging if you don't know where to begin analyzing.

That said, let's start by addressing the concrete steps that will narrow your search for actionable insights.

Accept the "Hard Truth"

You can't find courses that teach you how to derive actionable insights because the process largely depends on your business domain.

There's no cookie-cutter solution when it comes to data analysis. You'll first need an understanding of your industry and how your firm strategically fits into the market. You can assess your industry knowledge by asking yourself:

• What affects profitability in my industry?
• What are my firm's resources and capabilities, and how are we matching both to execute strategy?
• What are the specific problems my business is experiencing?

By asking yourself questions like the samples above, you'll better understand what data should have your attention, which will vary by industry. For example, valuable data in the financial services industry might not be helpful in pharmaceutical manufacturing. Understanding what data is beneficial requires industry knowledge, an aptitude developed on the job.

Although what is considered insightful varies, it's much easier to determine whether insights are actionable. Actionable insights will be something within a company's control to affect. Let me provide you examples of both non-actionable and actionable insights:

Not Actionable:

• Our monthly web visitors decreased by 5% this quarter.
• In a recent survey, 15% of our members mentioned membership benefits.
• Customer churn increased by 2% this quarter.

Actionable:

• People talk about our brand more negatively after a recent logo redesign.
• Customers are pleased that our business offers flexible membership options.
• 75% of churned customers said a competitor's product was easier to use.

Actionable and non-actionable insights may provide good-to-know information, but actionable insights go further by forcing you to rethink a situation and consider new solutions.

However, suppose you lack the industry knowledge to know what is considered insightful. In that case, you should start honing adjacent skills, particularly communication, to increase your business acumen, which leads to the next step.

Ask Questions

Behind every analytical problem is a business one.

To solve any business problem, you must first identify your firm's Current State (where your firm currently is with the problem) and its Desired Future State (where you want your firm to be after implementing the solution). What prevents your firm from moving from the Current State (CS) to the Desired Future State (FS) is your Gap. The Gap between the CS and FS is your analytical problem. We will refer to this problem-solving method as the CS-FS Framework.

Whether you're working internally or client-facing, it is beneficial to speak with your stakeholders to understand how they are experiencing the problem and what they are currently doing to alleviate it. You don't want to spend time developing a solution only to be told that it has already been tried. Listen to your stakeholders, find out what hasn't worked in the past, and ask questions as you listen. Some helpful questions might include:

• Can you describe the problem that you would hope to solve?
• What business decision would you like to make?
• How have you tried to alleviate the problem in the past?
• What inside resources do you have available, and are outside resources required?
• Do you already have an end vision for the project?
• Who would be the end-user of this analysis?

These are good starter questions that can help you pull information from your stakeholder to understand the Current State and picture the Future State. Being able to layout the CS-FS Framework fully will help you understand the project scope.

For example: Suppose after meeting with a client, you determine that your goal is to increase their "purchase rate" by 10 percent, (which would be the firm's Desired Future State).

You might think to look at existing "sales funnel data" to see where users drop off most before placing orders (which would be the firm's Current State).

At this point, the Current State and Future State are established and the Gap is defined.

Next, you would likely recommend some options to test that could reduce user dropoff. Let's say you recommended adding PayPal as a payment method to streamline the checkout process.

So, as you can see, knowing the CS and FS can dramatically narrow down your search for actionable insights.

Once you have identified the CS, FS, and the Gap, you will have your analytical problem and most likely will have a better idea of what data you are looking for to find actionable insights, which brings us to the next step.

Find the Analytical Solution to the Analytical Problem

Start by forming a hypothesis before you look at your data.

It's essential to establish your opinions first to avoid developing biases. A hypothesis should be independent of any available data. Ideally, you will form many hypotheses and eliminate them as you progress through your analysis. As you undergo testing, modeling, and other methods to analyze the data, cross off the hypotheses with insufficient data or any that lose reliability. In other words, solve the analytical problem.

Remember that this step's output is merely an analytical solution, and you will have to convert it into a real-world solution, which is the final step.

Convert the Analytical Solution to a Business Solution

There is a difference between being correct mathematically versus being correct rationally.

The reason is that analytical solutions are for computers, while business solutions are for humans. Since you are working with humans, you'll need to convey the fruits of your analysis in non-technical terms and ensure your final results make sense in the real world.

Think about it this way: What's the point of developing the most efficient and accurate machine learning model if your final stakeholder cannot understand the results. Despite how excellent a job you've done, the analysis is useless if the final stakeholder cannot derive meaning from it.

You can help your final stakeholder understand your analysis by creating storyboards that tell a story of how you went from the CS to FS (or, in other words, explain how you overcame the Gap). This part is where you implement your visualization, dashboard creation, and presentation skills to unveil the actionable insights you generated. However, understand that displaying a story is more than laying out a few colorful charts on Power BI or Tableau. Every chart and number presented should be telling a part of a larger story.

That's a wrap: That's how you derive actionable insights from start to finish.