Same process, different destinations
In physical sciences (biology, chemistry, etc.) after background research is done on a VERY specific field, a bold hypothesis is developed, testing conducted, results analyzed, and a conclusion is reached.
Typically, these hypotheses aren’t directed towards a particular application; the goal is simply the gain of information. Physical science is dedicated towards expanding the base of human knowledge, often times without knowing the exact application of that knowledge.
Because of the explosion in the last quarter century of our ability to COLLECT data, there are practically an infinite number of observations to be made using analytics. This has led to an approach where hypotheses are developed in data science AFTER testing is conducted.
While Data science SHOULD behave like the physical sciences regarding this process, it should be fundamentally different from physical sciences with its intentionality. Analytics teams often ask questions like “which of these variables are correlated” or “what makes these variables different.”
These are interesting questions, and with the mountains of data available, they can certainly be answered. There are countless stories of analysts who spend hours and hours developing brilliant insights and presenting them to business leaders, only to be met with a lack of enthusiasm and a “So what?”
2 questions behind ALL insights
The issue here is not with those business leaders. They are spot on. It is the responsibility and core function of an analytics process to provide the “So what?” by clearly articulating how their insights will answer one of two questions:
How will revenue be increased? How will costs be decreased?
It sounds simple, because it is. The reality is that increasing revenue and cutting costs is management’s number one priority, just about at all times. That company doesn’t have an analytics team to make “interesting” observations. Business leaders don’t want to have to figure out on their own how to utilize data insights in such a way it answers one or both of these questions. They have a thousand other tasks and responsibilities on their plate.
That does not necessarily mean that every insight has to provide improved cashflow immediately. More often than not, it requires several steps to go from a novel observation to putting a business process in place to take advantage of that knowledge and have it impact the bottom line. However, this DOES mean that the pathway should be clear.
Analytics does not function without buy-in from key stakeholders. Those decision makers will provide that support when they are confident that the data insights provided will help them do their job: boost revenue and cut costs.
Insights must make a clear impact
The retail end of the financial sector provides a prime example of this concept, where the key decision makers are external customers.
With the advent of easy-to-use online trading platforms, anyone with a smartphone can participate in trading the stock market from just about anywhere. With this ability comes the demand for more knowledge as to what’s going on with stocks, and what trading strategies are useful. There are organizations whose sole aim is to arm these retail investors will trading strategies they can use to consistently generate profit.
The better the data insights provided by these retail investor educators, the more value they are providing their customers, and thus the more profitable they will be. Therefore, these educators should have a laser-focus on providing insights that will make an IMPACT. Insights that clearly answer the question “So what?”
Many trading strategies marketed towards to retail investors claim to have an amazing “win rate.” They will point out that their strategy generates profit 90% of the time or higher, hoping to catch the eye of their customer.
In trading, a “win” is defined simply as having generated profit. A trade that generates one penny is a win; one that loses $10,000 is a loss. You can see where this is going.
Win rates should not be the primary insight delivered to potential customers. How does win rate address increased revenue or decreasing costs? It doesn’t. A customer could win with a trading strategy 99% of the time and lose boatloads of cash over time, assuming the loss was greater in magnitude than the profits put together.
How can these strategies be marketed better? “Trading strategy X will provide 28% return on investment.” Now the customer can see how their revenue will be increased with this strategy. They are more likely to dig deeper, and to strongly consider purchasing said strategy (if it exists).
If the 90% win-rate is mentioned in the context of consistency, that makes more sense. However, this is a secondary feature and only should be used if the educator knew that their customers put premium value on the consistency of the income. Even so, it is still less important than the RETURN the strategy will generate.
Know the opportunity cost of proposed changes
Solutions are rarely as simple as ‘take action X and make more money.’ The vast majority of the time, meaningful insights that address how revenues can be increased or costs slashed will point to specific changes that need to be made to existing operations and strategy. These changes will likely have a negative aspect somewhere in the business, even if it’s for the overall greater good. After all, there is a reason those processes were there in the first place.
These negative consequences must be accounted for when estimating the opportunity cost of an action. In case you’re not familiar with the concept, opportunity cost is a term in economics that addresses the value of the actions you will NOT be able to take by making a specific choice.
A real-world example is a question most young couples must answer: is it worth it to have a fancy wedding?
Many couples decide it’s important to them, others prefer to have a smaller wedding and use that capital that would be spent on extravagance on the down payment for a home. Neither one of these options are ‘correct’; it depends on the value the couple places on each. This is both a quantitative and a qualitative question (which is why the answer depends on the people involved).
The couple that uses the money on the down payment determine that the opportunity cost is too high for spending the money on the wedding; they find the value of having their own home greater than the value of having an unforgettable celebration with their family and friends.
If a couple believes the wedding experience is a life event they want to fully appreciate and that owning a home is nice but not a priority, they deem the opportunity cost of having an extravagant wedding is low enough, so they choose that option.
Let’s look at a high-level example of how this plays out in healthcare.
Patience for patients
Automation is becoming more and more prevalent in healthcare, particularly with administering pharmaceuticals to patients in hospitals. Different wings of a hospital house patients that have specific medical needs, and thus require certain types of pharmaceuticals.
The process of distributing the inventory of the hospital’s pharmacy across the appropriate wings was once based on the experience and estimation of the staff. Granted, these estimations were likely grounded in logic and reason and were GOOD, however they were not OPTIMAL.
With the incorporation of machine learning and artificial intelligence, pharmacists now have the ability to order and distribute drugs based on sophisticated analytics. Improving the distribution of drugs means they are more LIKELY to get to the right patients at the right time, thereby improving health outcomes.
Under the Affordable Care Act, a hospital’s bottom line is directly tied to the health outcomes of their patients. Therefore, this both increases revenue and cuts costs.
While this improved distribution is no doubt for the greater good, there is an opportunity cost for implementing this system. Not only is there the possibility of their being a technological bug in the system, redistributing drugs towards one group of patients means they will be pulled away from others. This is a natural opportunity cost of changing inventory patterns.
There is also a human opportunity cost. Medical staff that have worked their entire careers without AI will have to adjust to the new patterns. They have been accustomed to having specific drugs in one wing of the hospital, and the conventional wisdom may have changed.
On top of that, some medical staff may not trust that a computer’s logic will be better than their own. Hospitals must determine if utilizing AI is worth medical staff possibly feeling less empowered.
To sell your insights to management, be a salesperson
The takeaway here is that for any insight and improvement to be implemented, it must be supported by the key stakeholders. In other words, the conclusions from analytics must be sold to the internal customer.
What does our internal customer care about? Increasing revenue and cutting costs.
In the same way a salesperson handled objections before they arise, the same must be done when selling management on insights and proposed actions.
To go from an idea to actual improvement using analytics, it’s best to use the following process:
1. Develop a hypothesis that clearly leads to increasing revenue or cutting costs
NOTE – it’s perfectly ok to be incorrect, that’s a part of science. It’s usually not expensive to find something that doesn’t work. It’s far more expensive to think something works, implement it, and find later your assumptions were incorrect. Using Thomas Edison’s logic, you gained knowledge as to what is not effective.
2. Determine what change in operations would need to take place to lead towards an improved bottom line
3. Analyze the opportunity costs of making the proposed changes
4. Outline the specific next steps to either move closer to boosting the bottom line, or to discover the next insight that can lead to it
5. Articulate your findings to management
At this point, all management needs to do is give your proposal a thumbs up or a thumbs down. You likely won’t be making a proposal if it wasn’t clear from your research that it makes financial sense to move forward.
You want the decision maker to have as easy of a decision as possible. By handling their possible objections ahead of time, you won’t have to be facing a “So what?” after putting in all of your hard work.
And that’s the bottom line.