Fabrizio Salvador is known around the world as a leading expert in operations management. He is a professor of operations technology at one of the leading business schools in Europe. He has already helped companies such as DHL, IBM, Nokia and Unilever harness digital technologies to improve operational excellence and, having interacted with hundreds of high level manufacturing executives over the last few years regarding their transformation, he has come to recognise that, although it’s a tailored journey, there are phases that can be identified in the process: groundbreaking, experimenting, piloting and scaling. You can read Fabrizio’s latest article on the foundations of a digital transformation for more information on the subject.
Q: How firms need to reorganize internally to be successful in the digital economy?
If you really look at the best-in-class digitalization it’s not just about technology and it requires a deep change within an organization. There are at least four areas of the organisation that need to change. First, you need to remember what is the goal and it’s to create algorithms of data rules or decision report systems that help people make better decisions. How do I staff the maintenance team? How do I diagnose a problem? How do you know when a machine will need a replacement? Or how broad should the cross-training be of an employee on the shop floor? All these questions require, above all, a new mission for IT. The new mission is on top of routine tasks of creation, maintenance and ease of access to data lakes. They need to have new data engineering skills to accumulate data for operational decisions that can be fed to those developing new algorithms.
This is where the second function required for organisations comes in, namely the analytics development team. This is a team of people whose goal is to create new algorithms. It needs to be cross-functional as you need data scientists, process specialists, data engineers as well as lead users to check whether it makes sense for the end customer. Then you also need to think about where will the team be based? Whether locally or whether you have a central expert across different plants, partially internal, partially external. At this point, a lot of people forget about one fundamental thing – creating an algorithm doesn’t improve your KPIs. You need to embed the algorithm within the processes and people, their everyday work. If I want to create an algorithm that tests maintenance operator, how much time does he or she have to fix a machine before you incur oil losses, for example, where do I put the algorithm? How do you know? You have to interface it with the systems and these require process engineers, man-machine interface experts, involvement with the users and enterprise IT, as It goes back to the infrastructure of the plant. What is important in this team, that I call algorithm deployment, is very different from the algorithm development because the development team is the highly intellectual, data science, innovative people that create these algorithms, dealing with complex problems. But when you think about stuffing the deployment team, this is people that have different ambition and interest, they like to see things running. They don't like to create a good they like to see that people are using something and doing something bad.
Finally, the fourth part of the organization that requires updating is management. All the people who govern the company at different levels, high level or middle level even low level managers, have to learn new skills, a new way of organizing and they also have to be part of this cultural transformation to have the faith in the power of the solution.
To sum up, four different organisational changes are required when transitioning to the digital economy in manufacturing: in IT, data engineers, in development analytics development team, in deployment we have algorithm deployment team that make the algorithm available across the shop floor and finally, the management.
Q: What is the recommended path for a factory to transition to a Smart Factory? How do you define what goes first?
Prioritization is another very important question. Many times when I talk to managers, they mention the problem of prioritising different improvement initiatives towards digital. I would say that the first thing to keep in mind is where your problems lie. Don't be fooled by the technology. At the end of the day, the technologies exist to serve a purpose. The starting point is where are your pain points, where do you think important areas for improvement lie or to create a new digital business and then you think about prioritization. The question is where do I update that? When talking about decision based model fed by data, you have to check whether this model could work first of all.
The third thing that you want to consider when prioritising is where are the low-hanging fruits because you may have data for a certain algorithm but the algorithm itself might be pretty complicated to develop. The fourth thing to consider is scalability. Scalability of an algorithm start with the scalability of the problem. The plant-specific problem might be a good low-hanging fruit but might not be as scalable as a problem whose characteristics resemble the problems common to other plants and facilities.
Q: How do you prepare your people for this change?
The first thing to remember with preparing your people for change is that patience is key, It’s a gradual process that will inevitably take time. If you think about the first phase of change I mentioned before, groundbreaking, what I’ve noticed is that, when you move from a non-digitised process to a digital one, you just start collecting data, store it, people start recognising that data is an asset.
The second thing that allows you to prepare your people is that people believe when they see something come to effect. In this sense, before jumping to pilots, which can be sometimes risky and perhaps disappointing, which means they can become a sort of boomerang in terms of creation of a digital culture, instead go for short-term experiments that are meant to prove something e.g. time it takes to maintain a machine to be able to better schedule machine serving resources, which might prove that an algorithm was better than the planner by 85% two percent of the times in forecasting the time it took the company to fix a bug. It helps people see the impact in the real life, which generates trust. Finally, when you are able to scale the solution, the message you’re sending to your people is very strong, setting the new standard saying ‘this is the way we do business now.’
Above all, however, what I’ve noticed in other plants, such as big automotive manufacturers in Germany playing with digitalization on the assembly line, is that sometimes we underestimate what our workers know about digital. The point is really try to learn more about your people. For example, at Audi when setting up the digital assembly infrastructure in their plant, I discovered that people were sort of line support engineers in their free time. They could program in Python or HTML interfaces or other digital technologies, but nobody asked them. If you ask, you might discover a few surprises. The final thing that you have to consider, and this goes back to the first question that I was asked, is about the specificity of preparing people for this change.
IT are the Guardians, developers are the gatekeepers of data lakes you have the people that develop algorithms, which is not only scientists, and then you have the people who deploy algorithms, making sure they are available on the shop floor, in the offices, in the supply chain for everyday use. Management then makes sure that everything is running smoothly. When you think about how do you prepare people for the change, we have to ask ourselves which people. How do I prepare IT people? How do I prepare analytics development team, deployment? The question of preparing talent must also be split across based on the different functions that are affected by this change.
For more information on preparing your operations and people for the digital economy, click here.