Data analytics are an increasingly important aspect to businesses-- but it is among the least understood fields-- outside of the experts who deal with it on a day to day basis. Corporate leaders, CEOs, and other executives have mostly left data analysis to the data scientists-- and have not made a concerted effort to understand it. Every aspect of a company can be broken down and analyzed with analytics-- but what good is all this information if it is not well understood or put to good use? Sometimes it can seem overwhelming to leaders-- and they are hesitant to get involved in something they know very little about.
But as Florian Zettelmeyer-- a professor of marketing and faculty director of the program on data analytics at the Kellogg School of Management at Northwestern University argues, the most important skills in data analysis are not technical-- but rather they are thinking skills. Leaders excel at these types of abilities-- and are extremely adept at making decisions for their companies based on the information presented to them. When they understand the difference between good data and bad data-- and the process of how to get good data-- they can consistently make better decisions.
The best way to get good, accurate data is to start with a problem. Lay out what your company is trying to solve-- whether it’s how to make a better product or how to have better customer service. All data collection should be in the service of trying to solve an issue. Data should not just be collected, just to be collected. This does not serve a purpose and can lead to inaccurate results, and poor decisions will be made as a result. Data analytics should also not be a separate part of the business plan-- it should be integrated into all aspects of how the company works through its obstacles and problems. Whatever it is that the company chooses to measure or to analyze, the results that are gathered can only be useful if there is a well-articulated problem to be solved.
When companies do not have a specific issue to solve in mind-- they can get into trouble when using the data that is produced and given to them from a broad analysis. This data has no purpose, and it has been gathered with no larger picture in mind. Data collection will not offer the insights companies are looking for just because it is data-- the only thing to be gleaned from purposeless data is that it is information that has been gathered. Gathering a broad spectrum of data will not lead to breakthroughs unless it is specifically tweaked for the problem at hand. If a company gathers a large amount of browsing behavior of their customers-- for example-- that data does not offer any good insights or provide any new information to the company. If their browsing history is analyzed to help the company come up with better advertising targeting-- then that is a problem where analysis of this type of data can yield some fruitful results.
Managers are the ones who make the decisions on what problems need to be solved-- so why are they so far removed from data analysis? Instead, leaders should be involved daily in determining what exactly needs to be examined-- and what the data gathered could mean for future decisions on this issue. But in order to be more involved in the data process, managers should understand the data generation process first-- and what it means to have good and bad data. A lot of executives tend to defer to experts when it comes to analytics, which isn’t necessarily the best way forward.
If executives want to know the difference between good data and bad data, they must first understand how data is generated. Instead of letting the data experts take care of this process, managers should be aware that in order to make good decisions, they need the appropriate data. Since most decisions can be boiled down to choosing one method or one group over another, it would help if managers knew how the data they are comparing was acquired. This brings us back to the most fundamental part of data analysis-- what is the problem you are trying to solve?
If a marketing department is interested in whether or not their ad campaign worked-- it would be helpful to know which of their customers were aware of the product before the ad, and which were not. Perhaps the portion of customers were aware of the product because they had purchased it previously-- or had gotten an ad at some other time. This will skew the data because you just don’t know which customers were brought in by the ad, and which were already interested because of prior knowledge. This type of data is not helpful-- and unless there is a way to isolate those with prior exposure to the product, cannot solve the problem.
When the data generation process is analyzed further, it can help companies target the problem of reverse causality. If an organization is interested in learning whether or not to limit targeted emails, their data that they have will probably show that these types of emails are indeed effective-- and the more that customers receive, the more purchases they will make. However, if we look harder at the analytics, it will show that repeat customers are more likely to buy frequently and spend more per purchase-- generating more targeted emails in the process. So is it really the targeted emails themselves that are bringing in the revenue, or the purchases made that increase the targeted emails? The data is skewed once again.
Leaders have an enormous amount of domain knowledge of their industry-- whether it is what their typical customer base is comprised of, what products do better at certain times of year, or when is the best time to move inventory. They should tap into this knowledge and use it to their advantage. This is especially useful if their data analysis show a strange result-- one that is atypical of what a manager can normally expect within their industry. Leaders should take in account outside influences that can drive purchasers to them one month-- and away from them the next month. This data will not just offer up insights on its own-- which is why it needs the thinking skills of a business manager.
A leader who is aware of how the data was generated and knows about the data analysis process will be able to look at a data set and decide if it is good data or bad data-- and whether or not such data is useful to solving their issue at hand. Just having the data is no longer enough, and a business leader needs to be able to take their in depth knowledge of the industry to examine and uncover solutions in the data. Without thorough analysis applied towards solving a problem, data is just data.
Much in the same way that evidence-based medicine is leading the way to revolutionizing healthcare and treatment of patients, leaders who have a grasp on how data analytics can be employed to help solve problems and provide a better service or product to customers will be at an advantage. Having the initiative to get involved in this important process shows employees of the company that the manager or CEO is willing to get in and learn something new for the future growth of the company. This makes it evident that the executive levels are taking data analysis seriously, and believe that this type of analysis is the way forward.
With a commitment to using the right data to solve problems, the employees of the company should also be informed of how data is generated, as well as the types of questions which being asked to generate it. When the company itself is data literate, questioning the established processes are open and available to everyone. This input can be very valuable and lets the employees on the ground know that their insights are considered-- it gives them a stake in keeping the company innovative and flexible.
More and more it is becoming necessary for executives and other leaders to understand the methodology behind data analytics. If they are serious about leading their companies forward-- while stimulating innovation and growth-- they should be eager to learn how good data analysis can affect all areas of their business.