Groundbreaking ideas and research for engaged leaders
Rotman Insights Hub | University of Toronto - Rotman School of Management Groundbreaking ideas and research for engaged leaders
Rotman Insights Hub | University of Toronto - Rotman School of Management

Management analytics explained

Read time:

Opher Baron

How do you define management analytics?

Management analytics involves developing a precise understanding of the factors influencing managerial decisions, and designing analytical solutions to support those decisions. Put simply, it extracts insights from real-world data. A typical analytics project starts with acquiring a deep understanding of the managerial problem at hand, then identifying the data sources that are most relevant to it, creating data sets and designing and implementing analytical solutions.

In today’s world, every move people make online generates some form of data. The smart use of this data can help organizations generate competitive advantage, improve their customer experience, identify risks and optimize their operations. One of the main uses for analytics is around balancing supply and demand. For instance, figuring out what the demand will be for your product tomorrow or next month. While COVID-19 taught us that we can never know exactly what tomorrow will bring, putting today’s and yesterday’s data to work for a business definitely helps it keep supply and demand in balance. It’s not perfect, but we can predict things significantly better with data than without it.

Supply-chain management has been in the news a lot lately. How would you describe the current situation?

Our supply chain has been impacted by everything from global ports being closed for extended periods to manufacturers and food producers shutting down. Given all of that, it has actually worked pretty well; but I’m not sure it can withstand much more. The global supply chain has been a bit like a boxer who has taken some severe punches but is somehow still standing. It continues to be hit again and again by things like the flooding that closed the port of Vancouver recently.

When you add inflated gas prices to the picture, it is clear the supply chain is very stressed right now. The first order of impact of high gas prices is obvious: "Filling up my tank will be more expensive." But the second-order impact is that essentially everything we purchase has to be transported somehow. If you buy something via e-commerce, a truck delivers it to your house; if you shop for groceries, multiple trucks will have delivered the products you purchase. All of these transactions have become a lot more expensive, and there will definitely be trickledown effects. I am a bit concerned about what we might see going forward — especially with inflation at 40-year highs.

In a recent paper, you discuss four distinct types of data analytics that organizations should embrace. Please summarize them.

Traditionally, descriptive analytics has been used on well-stated, existing problems. But modern data analytics enables us to find hidden problems within an organization — and to learn from best practices. Descriptive analytics is still the first level of the framework, because it describes the past performance of an existing system, but organizations should not stop there. Predictive analytics predicts future performance for existing systems; comparative analytics compares the performance of systems under different interventions; and prescriptive analytics prescribes interventions to improve future  performance.

Consider the phases involved in fixing a congested service system. The descriptive analytics phase would examine the system’s past performance and address questions like, Is our process performing as designed? and Are there bottlenecks that need to be addressed? For example, in a hospital emergency department (ED), this could lead to observations such as, "demand for our services has been highest on Mondays from 4 p.m. until 8 p.m." or "chest pain has been our most common patient complaint." The predictive analytics phase would consider the future performance of the system and address questions like, "What throughput can we expect?" and "When might a staff member be absent from their shift?"

The third type is comparative analytics. Here, you consider how specific changes to your systems would have impacted (retrospective) or will impact (prospective) performance measures. One approach to comparative analytics involves benchmarking data against other organizations in your line of work. Benchmarking is particularly important for organizations who operate several branches. Modern statistical methods support comparative analytics and help answer questions like, How would our system have performed if Improvement Project A had been implemented? In an ED, the team might consider the impact of various new workflows and staffing changes. For example, maybe the team will find that they should add a dedicated track for patients with chest pain. This type of analytics allows you to differentiate between your existing system and your system under different possible interventions.

The final type of analytics is the most valuable one. Prescriptive analytics considers how to best manage your future performance. In this phase you can address questions such as, What is the best staffing procedure for us to follow? Or What is the best approach to customer routing? In an ED, this final phase might search for the optimal nurse-doctor ratio at different times of day and lead to shifting schedules or guidelines for when to call in extra personnel. The prescriptive phase can only work well if it includes the knowledge obtained from the descriptive, predictive and comparative analytics phases.

You believe comparative analytics is particularly important for service-based organizations. Why is that?

One of the most important inputs into the effective management of any operation is the cost-benefit analysis of different interventions. The most common interventions in services are related to changes in capacity (e.g. moving shifts, adding capacity or changing processing times or processes. Evaluating the impact of such interventions is a powerful tool.

People tend to use data in a very linear way. They might look at their descriptive data and find that when they priced their product at $100, they only sold 30 units, but when they priced it at $80, they sold 100 units. As a result, they will move forward believing that this price results in this outcome. But the demand variation could have been tied to other things. Comparative analytics tries to understand the cause of the outcome. For example, maybe seasonal demand was extremely low when the product was offered at $100, and the price had very little to do with low sales.

How is AI and machine learning affecting all of this?

As indicated, we tend to think relatively linearly, but AI and machine learning enable us to make predictions that are non-linear and that consider multiple factors. As a result, they enable us to better predict tomorrow using data from today.

The Holy Grail of data analytics is pinpointing a novel intervention that improves a system’s  performance. Does this require all four types of analytics?

If you go through the entire framework — covering all four — you will have an excellent understanding of how the various elements of your system are working, and how they could work better. One common issue in today’s organizations is that the people who know how to work with data don’t know the business all that well, and the people who know the business really well don’t know how to work with data. The trick is to pair solid data analysts with the people leading your business. That’s the best way to not only improve understanding of your current system, but figure out how to increase its effectiveness.


How do consumers benefit when all of this is working well?

The easiest example is recommendation systems. When you go to Netflix, you see that because you enjoyed Action Movie Number 1, they are highly recommending Action Movie Number 2. This is done with an AI algorithm and it is one small way that data analytics improves our lives.

Another example is that when we go online, we all see different ads on our screens. Like the movie recommendations, these have been tailored to our interests. The algorithm somehow knows that you are thinking about travelling abroad this summer and that in the past, you travelled to France. So it might serve up some tantalizing photos of Paris or Normandy and provide a quick link to purchase a plane ticket. This customized advertising not only allows you to save time; if done well, it can change your behaviour in ways that improve your quality of life — while at the same time, increasing profits for the company involved.

Which industries stand to benefit the most from management analytics?

All industries can benefit, but there are two in particular that stand out for me. The first is autonomous vehicles, in the context of ‘how to run a car safely. ‘The computing power in these vehicles has to know how to operate in different weather conditions, different traffic situations and so on.

The second area is one we’ve already touched on: healthcare. I am a huge proponent of using more management analytics in our healthcare system. The idea is, again, that we can use the copious amounts of data out there to improve the delivery process. Doctors are great, but they are also human, which means their memories are limited and they suffer from biases like the rest of us. If we can provide them with information that is based on large amounts of data from other patients, they will be able to make better decisions. For example, millions of people have chronic diseases like diabetes, and lots of them have a health profile that is very similar to mine. The data could show, for instance, whether people like me respond well to a certain medicine. We can in principle provide this type of information to doctors in real time.

Any analytics exercise depends upon robust data. Do you feel like the data being collected is up to the task?

There’s an old saying about data: garbage in, garbage out. You give me garbage data and I’ll give you garbage output. I think more and more companies understand this and recognize that creating value in the future will largely depend upon data analytics. Many companies are taking steps to improve their data collection to make it less garbage-in, and that is great to see.

What would it take to future-proof the global supply chain?

COVID-19 taught us some critical lessons about supply chain management and managing risk. Clearly, there are things companies should be doing better. For example, they need to have some production capacity close to home as well as more flexible production capacity elsewhere. In a world with pandemics and natural disasters, this is critical. The second big lesson is that things need to keep moving no matter what, in order to ensure people’s well-being. And going forward, embracing insights from data and management analytics will be critical to achieving that.

This article first appeared in the Fall 2022 issue of Rotman Management magazine. Published in January, May and September, each issue features thought-provoking insights and problem-solving tools from leading global researchers and management practitioners. Subscribe Today

Opher Baron is a distinguished professor of operations management and academic director of the master of management analytics program at the Rotman School of Management.