5 ways to get more output from the same assets (hint: listen to the data)

Schubergphilis website portret Marcelvan Ruijven
Marcel van Ruijven
Feb 08, 2022 · 6 min lezen Engels
Schuberg Philis factory

Machines have long been producing data, and quality data has long been available. Some would say it’s been around since the 1950s – because machines’ programmable logic controllers are fundamentally the same as they were decades ago – but the data was rarely collected. Today, we know how to utilize it and make it work in our favor. Plus, because cloud technology has become mainstream, ubiquitous, and affordable, now we can get even more out of data.

So how do we do that? At Schuberg Philis, we maintain that we want to understand the problem behind the problem. By that we mean: yes, we’re an IT company, so of course customers come to us with IT questions, but how we respond to them is not purely by solving a tech problem. Instead, we create data solutions to address the business challenge that underpins the IT issue. To achieve this, we keep five principles in mind, which we explain here and illustrate with examples from a recent customer case.

1) Start with the end user

Diving into a new project, we start with the problem. That refers not to some problem statement a manager has come up with, but rather the end user’s actual problem. Relatedly, we don’t take the initially reported problem statement at face value. We go straight to the source of the problem – whether that’s the customer’s workplace, field site, or factory – and the end users who are experiencing this problem on a day-to-day basis.

When a large beverage company asked us to help increase performance KPI, our team travelled to one of its main factories. Once there, we met the operators who were in charge of the machines responsible for the performance KPI figures. First we had lunch together. Then we accompanied them to the factory floor to see the real-life issues happening in real time. As data scientists, we could have presented the sophisticated AI models and machine learning solutions that we like to keep in our back pockets, but we didn’t. We had to ensure that the problem the managers initially reported to us captured what the operators were experiencing on the factory floor.

2) Look and listen for ourselves

To really understand what is behind an IT question and to propose the best, most fitting data solution, we need to look and listen for ourselves. We need to understand both specifically what the end user is experiencing and, more generally, how this problem fits into – or rather, thwarts – the company’s business goals.

In this customer case, we witnessed an operator distressed by the periodic stoppages that the factory machines kept experiencing. While on the floor with him, when a stoppage occurred, we saw the operator run around the giant factory hall, trying to determine which machine had stopped by scanning the eight enormous machines’ screens. Once he identified which machine caused the stoppage, he ran to a closet, grabbed a broom, and used it to clear out a stuck bottle that had caused the stoppage. And what did we hear? The sound of crashing glass, which had actually preceded the scene with the broom. One machine’s stoppage caused a chain reaction on the factory line since all the machines were connected.

3) Understand the context

Besides understanding our customer’s IT needs and greater business goals, another major component of providing the best, most fitting data solution requires knowing the context in which a problem exists. That requires getting up close and personal with our customer’s employees and understanding the culture and corresponding behaviors of the organization.

When we visited the beverage factory and met staff, we also came to learn that there were two broad categories of operators, each with a distinct way of working. More senior operators relied on their ears; they were alerted to a slowdown on the packaging line by the sound of it. Meanwhile, the younger generation of operators was more visual; they grew up with smart phones, so it was instinctive to rely on information shown on handheld tablets. The solution we eventually helped implement didn’t force either category to give up their habits. It still let the analog group use their ears, while supplying data-rich dashboards for the more digitally inclined. What’s more, in getting to know the people in the company, we were able to identify candidates who could serve as data stewards. Already data savvy, these individuals were ideal trainees for us because they could then train their colleagues, develop training materials, and roll out new solutions to share with operators.

4) Create solutions in tandem

Our strong engineering background – with a 100% uptime guarantee for mission-critical activities – has a halo effect on all our solutions. As a result, our data engineers ensure that any data solution has a solid foundation. Concretely speaking, this involves effective data governance, meeting security and compliance regulations, and effectively taking advantage of all the cloud has to offer. But we also combine that with our business-mindedness, insisting that IT solutions should work in tandem so customers get value out of data solutions.

As an example, changeovers on the production line at the beverage factory were sometimes preceding machine maintenance time by two hours. This led to frequently having to start and stop the production, wasting operators’ time and thus decreasing the performance KPI. Utilizing data gave us a bigger picture of the trends, and we proposed a solution that would combine the switchover and the maintenance into one step.

5) Stay in our lane

Any solution we develop is done in close cooperation with our customer. We often say that we put experts in the lead. The experts are our own colleagues and those of our customers. We also know when we don’t have the expertise on a certain topic or for a particular field; we know to stay in our lane. For as much as we have technical knowledge, our customers have the domain knowledge.

Working with the beverage company, we came with data science and engineering expertise, but the operational experts were the factory operators themselves. They had years of hands-on experience and an understanding of the production line that we could not have. Our solution therefore involved utilizing the data that the machines were already producing and connecting the data points across all the machines to create robust insights. These insights provided trend analyses and gave the operators a way to predict or prevent stoppages. In tangible terms: our solution provided a self-service dashboard that operators could tailor according to their own needs, which they understood best, as well as a managing platform to which they could keep adding new data points.

Also worth mentioning is that the data being collected was the company’s own data and remained its own. Staying in our lane also means we don’t sell data back to its rightful owners. Customers’ data is theirs to keep. Our role is to help companies better utilize this valuable asset, thereby decreasing disruptions, increasing efficiency, and thus getting more output. In short, we get them listening to the data.

By Bastiaan Timmerman and Marcel van Ruijven