COMPANY CONFIDENTIAL. DO NOT DISCLOSE.
Günes Pekmezci of Bundesdruckerei shares advice on securing broad support for analytics transformation initiatives
Günes Pekmezci is a manufacturing engineer at Berlin-based Bundesdruckerei, a state-owned security printing house responsible for the manufacture of official German passports, ID cards, banknotes, stamps, visas and vehicle documents. As technical project manager spearheading the implementation of the organization’s IoT technologies, Günes has fostered transparency around the adoption of new, more data-driven automated systems. She joined Bundesdruckerei in 2016 after completing an MS in Process, Energy and Environmental Systems Engineering at the Technische Universität Berlin.
Meg: You began your career at a time when Industry 4.0 was first gaining significant momentum in Europe. I imagine that, as someone coming out of graduate school already with an IoT interest, there were a lot of exciting opportunities to drive change. How would you say this focus on analytics transformation shaped your career trajectory since joining Bundesdruckerei?
Günes: Coming into Bundesdruckerei, I was very eager to optimize production processes, but knew little about the new industrial revolution. I started as a trainee in a division of the production department working with one colleague to develop Industry 4.0 initiatives.
I began reading and researching this new revolution and ultimately created a maturity system for our machines, related business cases and portfolio with a road map. My manager and I were the only two in this department at first! She became my mentor and supported my creative process, encouraged me to learn more about the subject and gave me valuable insight into business use cases. We were very successful with this approach, so our maturity tool was even used for future projects.
During this research period, I was working to optimize Bundesdruckerei’s production processes with data – an effort I ultimately brought to the engineering department as well.
The transition totally changed my career orientation – from chemical engineering to data engineering and data science – and I became the chief architect of our data streams, seeking ways to better collect the data needed for analysis. I evaluated a number of different IT systems and worked closely with our Production IT group to implement the systems we procured, eventually leading to the Industry 4.0 infrastructure we have today. I was very hands-on during this process as I tested all IT Programs for the use cases that we needed. I now see myself more like a self-taught IoT/ IT engineer.
Meg: What were some of the hurdles you had to overcome in implementing these new systems and data processes?
Günes: For one thing, Excel was widely considered the preferred tool for data analysis at Bundesdruckerei. In the beginning, it was hard to show that other tools could do better… hard to convince people to try something new.
My strategy was to understand their needs and to show that them how they could fulfill their needs faster and more easily with JMP. After seeing how fast JMP could run an analysis, one of our older colleagues even started taking the time to learn this new reporting tool and she’s now one of the best JSL programmers on our team. She knows how to automate her own reports with JSL scripts. It is so valuable to see even longtime Excel users find joy in using JMP instead.
Meg: Introducing new methods or tools can definitely be disruptive at first! How were you able to bring the skeptics on board with your vision?
Günes: At the point when I first concluded that we should be using JMP to analyze our data, there was one team that had been using Excel macros for a long time and raised some objections. My approach was to understand their reports and build the exact same analysis in JMP with automated scripts for a direct comparison.
And we ended up reducing analysis time by 90%. In Excel, it took 107 minutes to process 30 weeks’ worth of – and Excel reached its limit and could not handle more – whereas in JMP I showed we could analyze a whole year’s worth of data in 13 minutes!
Therefore, they saw the potential and became less critical. I sat with them and showed how they could implement their own scripts. With their satisfaction, I rolled out JMP with this promise of ease and time savings.
I’ve also organized trainings and created weekly JMP office hours during which colleagues can come to me for assistance finding a solution in JMP for a specific an analysis or data challenge. It’s very hands-on. And I’ve found that the more people who have joined in, the more they have taught others. Some have even started learning on their own more and more and we have many JMP experts in our department right now – it’s something I am very proud of.
And it has been well received by our leadership. We’ve shown them that we can save time and energy for the same things we’re doing in other tools. JMP being very user-friendly has also helped engineers start using it faster without being too scary. Our aim was to find a tool that didn’t require programming language knowledge since we want engineers to spend their time analyzing data rather than learning to code.
Meg: I’m curious – how has having JMP changed the relationship between statistical experts and domain experts at Bundesdruckerei?
Günes: I’ve learned to communicate carefully around implementing digitalization use cases in production for exactly this reason. I always underline the fact that digitalization will support domain experts’ expertise. And while it may take a load of repeated work off their plates, it cannot work without that domain expertise.
Machine learning requires a lot of expert decision making in the algorithm training phase and at the end, once the machine starts making decisions, it can only do so much to guide you. As humans with complex brains, we see things in the data that a machine may not be aware of – like physical conditions, for example. The expert will be the one who will decide to greenlight a decision in the end. These new revolutions are there to make our lives easier.
Meg: What has your relationship with the JMP organization – account managers, technical engineers, customer advocates – been like?
Günes: I feel extremely lucky to have started working with Jessica Horsch, the JMP account manager. She is extremely determined, and we have solved so many autocratic issues with the help of her determination and positive thinking. She even managed to get a contract signed shortly before New Year when in Germany, almost no one works after second week of December!
On top of it, through her, we met Florian Vogt and Dieter Pisot. So far, we’ve done a lot with Florian. He is our preferred JMP Expert! We organize a few workshops on specific themes throughout the year and are very grateful to be able to “ask Florian” and have him teach us the specific things we need for our analysis.
Meg: What advice would you give to someone who is just starting out with JMP – or maybe looking to engage their organization in analytics transformation as you have done?
Günes: When I was first starting out with JMP, I watched a lot of web sessions – and used it in a hands-on way. I am also a fan of the scripting index in the tool. I Google every question I have and there was a time in the beginning when I read the JMP Scripting book by skimming from start to end. It was a guilty pleasure that I would only advise for geeks like myself!
As for culture change, do not underestimate anyone’s expertise. Try to understand their needs and reasons for why they use a specific system – and be respectful about their existing systems. Create the change with them, start with the things they feel comfortable with and guide them toward a new culture. Their acceptance is important in order that the new culture may evolve to become the standard. And it’s critical to involve everyone with a relevant role from the very beginning. They shouldn’t get the change dictated to them from the top down; rather, they need to be part of the transformation.
