How Predictive Analytics is Driving Supply Chain Resilience

Igor Rikalo is President and Chief Operating Officer of o9, having joined the company almost 14 years ago. He has witnessed the company grow into a global player, with more than 3,000 employees and hundreds of clients around the world.

“I joined the company when we were just a handful of people with a vision,” he says “and getting into the execution mode over those 13 and a half years has been a really rewarding and humbling experience to where we are today.”

Igor has been a part of o9’s global scaling, having built high-performing teams in order to develop strong solutions for o9’s customers. The company works to drive more intelligent planning and decision-making, creating better financial results and more transparency throughout the supply chain.

What is the role of predictive analytics within enterprise intelligence?

That is a key question, and a very pertinent one, especially in a world of constant change, constant volatility, constant complexity that we are facing. Some of these disruptive forces on an enterprise - from the outside of the enterprise - have been quite impactful, starting from COVID, then to a lot of economic disruption, supply chain disruptions due to other disruptions worldwide. 

We are now operating in a world where a lot of these large supply chains are very long, there’s a lot of different points of disruption that are possible, which makes the domain or the function of predictive analytics ever more so important. When we started o9, we always distilled it to the statement that we are helping companies answer three W questions: 

• what is happening and why in my business
• what is going to happen
• what actions should I be taking?

Predictive analytics as a domain went through a significant maturation over the last several decades from being a rather simplistic, naive kind of a forecasting, to using a lot more a statistical approach. This has transformed in the last 10 to 15 years, using some of the machine learning, advanced algorithms and actually evaluating the impact that each of those drivers can make onto the forecast to increase the accuracy of that prediction and reduce any bias.

Why do you think enterprise intelligence is important to supply chain resilience?

You can think about the enterprise in terms of being a people process. 

Technology is one traditional way where we look at the enterprise or what the enterprise does to serve their customers. Another way to look at it is to look at the enterprise as a series of decisions to be made and some decisions are very tactical. 

Where do I send this truck tomorrow? In which stores do I put my product in from this distribution centre? Some of them might be a little bit a medium horizon in a sense - what are the promotional campaigns that I should be running against this product line in these channels, over the next six to nine months? Some of the very long-term strategic decisions, like where should I put the next distribution centre?

That decision intelligence is what helps companies grow and have that growth, being effective and efficient, which is what is really important these days. And a lot of these decisions are driven by the models or should be driven by the models, which is what our perspective from o9 is. These decisions are fact-based versus a more traditional gut base that the management is trained to do.

How can companies learn from systems like predictive analytics and enterprise intelligence?

I think actually I would flip it. I think the objective is for those systems to learn the knowledge that’s in people’s heads because we don’t want to be pretending that these systems are intelligent or smart. They’re algorithmic representations of decision-making trees. So I would basically say that the objective is for the companies to digitise their knowledge inside of these models, and we’re now having that technology that can make that quite robust. 

I think we are at a better place where we can convert the knowledge that resides in people’s heads, to be actually encapsulated into the models. And then those models can replace some of the more simple decision making. Humans deal with ambiguity a lot better than computers. Computers have to resolve the ambiguity to black and white, one and zero, and some of the newer models are able to deal with the ambiguity a little bit better, but still I think the humans are superior there.

Can you share a client-based study where o9 has aided in enterprise intelligence?

We are seeing a huge amount of value derived from getting good at a few things. And it’s all about the forecast accuracy and what we are seeing specifically in the consumer packaged food and beverage space, which can get into the mid eighties to low nineties, which is quite remarkable achievement for the machine to be able to be so accurate with a very little bias. 

We’re seeing some of our clients achieving upwards of 80-85% touchless metric, which means that the humans are then focusing on a very small subset of exceptions, where they typically have either more updated data or more knowledge to handle those. 

Some of our clients are seeing a quite significant reduction in inventory that’s held in the supply network because the forecast signal is more accurate, it’s more stable, so you don’t have to have a lot of inventory buffers in your network to guard against the volatile demand signal.

What do you think the future of supply chain strategy looks like?

I think what we’re going to see is a big push for companies to digitise a lot of their supply chain planning processes and knowledge into these models to gain efficiencies and productivity on one end, but also to democratise some of these processes inside of their organisations. 

Right now most of the companies, their planning organisations are first of all highly skilled analytical organisations doing a lot of number crunching, but I think the way those are then propagated are through a series manually powered processes. 

Typically, companies are going to have something called sales and operations planning, cadence or some field processes to disperse what the central planning organisation does into the field for the execution. With the advent of some of the chatbot interfaces where you can get a very precise, well-reasoned answer, it’s going to allow the central planning organisations to communicate a lot better with the rest of the organisation.

So I see that as a major step change in how the value is going to get unlocked in the enterprise of the future. We call that an APEX operating model, which is an agile and adaptive planning and execution process and the management system, and I think that’s what we believe it’s going to be the main driver of the next level of excellence in this domain in many enterprises.