S&OP implementation lessons: AI forecasting before planners, right-sized maturity, and when to use consultants, with Johan Van der Straeten

About Johan Van der Straeten

Group supply chain and procurement manager at Reynaers Aluminium, the Belgian manufacturer of aluminium systems for windows, doors and facades. Before that, nine years in strategy and operations consulting at PwC, where he wrote Reynaers' operations strategy, and seven years at Philips Lighting as quality and production engineer, Lean Six Sigma Black Belt, demand planner for emerging regions and supply planner. Listen to the episode.

Industry, consulting, and back

Johan Van der Straeten is one of the few guests who has sat on both sides of the table twice. He joined Philips Lighting straight from engineering school on purpose, "you can only move into consulting when you have experienced the thing your clients experience yourself", spent four years in execution, then became a Lean Six Sigma Black Belt, Philips' internal consultant, and got his first planning projects that way.

That shop-floor start, he is sure, made him a better planner. "Whatever you decide in planning has a consequence on your shop floor, on your machines, on your output, on your shift regimes. It gives you the end-to-end view of the whole supply chain." He has since covered every box of the SCOR model, "not a specialist in all of them, but I have the right understanding of the base principles".

What S&OP looks like when it goes administrative

At Philips, S&OP was mature and had been running for years. Johan was demand planner for the emerging regions, where history forecasts badly and the process part, collaboration with sales and marketing, matters most. He also saw the failure mode of a well-established process. "If you don't set it up right, it becomes a talking forum. If you don't assign the right responsibilities, the planner is just consolidating inputs, making a proposal, and that's that."

"I spent one week a month, the first week, creating PowerPoints, capturing data, putting it in the right format, and then had a one-hour discussion with the market. It was a bit overdone." Six months as supply planner, week two of the cycle, showed him the connection: "If you don't do proper forecasting, you can never make a good supply plan, your inventory levels will be wrong, and you go back to this continuous loop."

When to use consultants, from someone who was one

Ben asked for his view on what a company should do itself and what it should buy. "I am 100% convinced that a company should use consultants, because as a company you have the responsibility to bring in fresh eyes and new perspectives. That's how you bring in what's happening outside your four walls."

With a condition. "Don't use consultants if you can do it on your own, especially for the motivation of people. If it's straightforward what has to be done and your people brought up the ideas, give them the empowerment and the time." The other legitimate reason is capacity: when nobody has the time, "the consultant becomes the enrichment of your workload".

And a warning about what to expect. "They will never be detailed enough to give you the full plan with all details. Don't expect that. But they bring more conceptual thinking. They take distance from the daily mess and look at things from a different angle." Benchmarks and visibility across companies are the other thing consulting gave him: as PwC's Industry 4.0 lead he saw "what was working and what was more the marketing story", which "you can only have if you have visibility on different companies".

Eating his own slideshow

Johan joined Reynaers to implement the operations strategy he had written as its consultant, covering footprint, capacity, processes and systems, just after COVID had made construction "Valhalla" and operations unable to keep up. "I stay connected to the real world by the challenge of implementing that slideshow. And I can say not everything I put on a slide has become reality."

The lesson is about time. "When you are on the consulting side, you forget there are so many things that happen in daily business that you cannot neglect. The plan was over-ambitious. I still think we set the right things; only the estimation of the timeline was too ambitious."

Starting demand planning with AI instead of demand planners

The strategy called for S&OP, which needs demand planning, supply planning and capacity management, none of which existed at Reynaers "in a structural way". Building a demand planning team from scratch looked like a year: six months to hire, six to onboard. So Johan skipped it. "Let's forget about the demand planners. Let's use AI software to build that forecast. It's a black box, let's only use the outcome and the typical process of forecast accuracy to see if we are good enough."

"We did that. It worked, actually quite well. But what I forgot was the people component: how to interpret the results, how to trigger the right conversations with your salespeople. If they are used to talking euros, how do you translate that to volumes? How do you tell salespeople they need to consider the mix?" In Reynaers' market, the architect draws the house and specifies the window; there is no push model. "That whole process is what I underestimated, and I think it was a lack of sector knowledge from my side."

Would he do it differently? "What I don't regret is that I jumped to action. If you have an idea, a vision behind it and a good business case, Reynaers does it. Did I take the right action at the beginning, knowing what I know today? No. I should have spent more time understanding the sales part of the process, and reflecting on how we were going to use it."

You may need less S&OP than the textbook

His second-order lesson is about scope. "I started from this concept: I know how S&OP works, demand planning at that level of detail, supply planning, the capacity plan. When I look at it today, maybe we don't need it in the way I envisioned. It was too much for what we need." S&OP never ends, "but there is some top point where, with the efforts I have done, I get the maximum output. Beyond that it's marginal gains. I overestimated what it would take to get to the top point."

Analysts become planners

A few months after the AI tool, Johan reshaped the supply chain analyst roles into S&OP planner roles. The distinction is his own. "I don't see an analyst the same way as an S&OP planner. I don't want people just to analyse things; they also have a responsibility for results and outcomes. Don't analyse just to analyse."

The skill shift runs "from somebody very good in Excel and data manipulation to somebody with business knowledge who knows how things are connected", plus influence: "You need more than the truth. You need to be able to influence people with the right arguments." What he wants from a planner after an S&OP meeting is ownership: not "somebody overruled me and that's that", but "we had a discussion, this is the best for the company, and I'm fully behind it." He is clear that it is an evolution, supported by him, and that process, systems and people have to mature together: "We made huge progress in certain areas and in some areas didn't make the progress we should. Now we need all these components at the right level to take a step up."

Replacing a legacy tool is harder than adding a new one

The AI forecasting tool was easy to introduce "because there was nothing existing". Detailed scheduling is the opposite. Reynaers has a strong legacy system, and "when you enter a field where there is already a legacy solution, you enter this discussion: why is our existing tool not good enough, can we not upgrade it?" A year of benchmarking and company visits followed.

The argument that moved the discussion was not technical. "Can you imagine a world where you have this option and that option? What will we do with that and how will it improve our process?" It started from a business goal, reduce lead times, worked back to what better planning would need, and only then to whether the legacy system could provide it. The fear is familiar: "There is an existing base and don't touch it, because if you touch this, the whole house can collapse. That fear can be right."

Asked where planning systems still have the most to gain, his answer is modest and pointed. "How many Excels are we still using in planning? There are so many solutions today to avoid that. But it can become such a complex IT architecture that people start to have their off-grid Excels." Self-steering supply chains he hopes to see, "but planning remains a people business. The tool alone will not create a solution. It has to be part of a bigger vision with processes, organisation and the tools that connect to that."

Procurement, suppliers and sustainability

Johan also runs procurement, which matters for an assemble-to-order business that is "very reliant on suppliers to have the right lead times". Collaborative forecasting and capacity alignment with suppliers are "very premature" today and his focus for the coming years. On sustainability, more than 95% of Reynaers' footprint is scope three, so sourcing decides it. What he is proud of: "Sustainability has become a parameter as important as all the other parameters you use to judge suppliers: total cost of ownership, service, quality, risk. We put it at the same level." For aluminium, the biggest contributor, that evaluation has been done "to the maximum level of detail", balancing recycled content against cost, service and quality rather than optimising one.

Your next steps

• Before hiring consultants, ask whether your own people have the idea and the time; if they do, give them the mandate instead.
• If you start demand planning with a forecasting tool rather than a team, plan the people side first: who interprets the forecast, and how sales will discuss volume and mix.
• Decide the maturity level you actually need from S&OP before designing the textbook version.
• Rewrite analyst roles as planner roles with accountability for outcomes, and train for influence as well as Excel.
• For a legacy system replacement, start from a business goal and ask the business to imagine the options before IT compares tools.

Johan's parting advice is to leave the building. At Philips he never went to events, "nobody knows it better than you, that was how I was raised"; consulting taught him otherwise. "There is a world outside your own company, and every event is an opportunity to learn." For the reluctant networker, his technique is simple: "Use the topic. Go to a table and say, what did you think about this topic? You can never embarrass yourself with that. Challenge yourself to at least two tables, and accept that some conversations are done after three sentences." And if you still don't know whom to talk to: "Go to one of the presenters and say, nice presentation, can you elaborate?"