An interview with Dr. Norman Dziengel (Head of Product) and Thomas Miller (Chief Customer Officer), Inpixon

Energy and critical-equipment manufacturers — turbine, transformer and compressor OEMs — are carrying order books they have not seen in decades.  Gas turbine backlogs at the major OEMs now stretch to 2030. Standard power transformer lead times average 128 weeks.

For plants building engineer-to-order equipment, where a single unit can represent 15 to 25 million euros in revenue, the pressure to deliver on time has never been more direct. Miss a ship date and you do not just absorb a penalty. You put your slot in the next procurement cycle at risk, at exactly the moment when winning that slot is worth more than ever.

I spoke with Thomas Miller, Chief Customer Officer at Inpixon, and Norman Dziengel, Head of Product, about where OTD actually breaks down in complex equipment manufacturing, what the shop floor has to do with a promise made in a sales meeting six months earlier, and why the gap between ERP and physical reality is the most expensive thing most plant directors are not measuring.

"In this market, late delivery does not just cost you a penalty. It costs you the next order. And the next order is worth more than any penalty clause your customer could invoke."
— Thomas Miller, Chief Customer Officer, Inpixon

When an energy equipment manufacturer misses a delivery date, where does the conversation usually start?

Thomas Miller: It usually starts in the wrong place. The post-mortem goes straight to supply chain, a late component from a supplier, a procurement delay, an import issue. And sometimes that is the real cause. But in a surprisingly large number of cases, everything needed to complete the unit was on-site well before the promised ship date. The delay happened on the floor itself: a sub-assembly parked in the wrong staging area, a kit that was 98% complete but nobody knew what was missing or where the missing part was, a unit sitting in rework with no visibility into how long it had been there. The commercial team made a commitment based on the production plan. The production plan had no real-time connection to what was actually happening.

Norman Dziengel: What Thomas describes is structurally predictable. ETO and MTO production, which is how all of our target customers build, means every order has a unique routing path through the facility. There is no assembly line where everything moves in sequence and you can predict position from time elapsed.

A turbine rotor arrives by crane, waits at a station, moves to the next, waits again. Between those moves, nobody has automated knowledge of where it is. The ERP shows you the plan. It shows you the last manual scan. It does not show you what is actually on the floor right now.

What is the specific moment when a delivery promise starts to break down?

Thomas Miller: It is rarely a single moment. It is an accumulation of small invisible losses. A sub-assembly finishes three days early, but no one in planning knows, so it sits. A kit arrives in the staging area but with two parts missing, nobody flags it until the technician shows up to start assembly and the work stops. A unit completes its test cycle and is ready for final prep, but it is sitting behind three other units in a queue that nobody has visibility into. Each of these is a delay of hours, sometimes days. In complex equipment manufacturing, where a single unit can be worth 15 to 25 million euros, those hours add up to the difference between on-time and late.

Norman Dziengel: The pattern I see across facilities is what I would call the staging area problem. In a large plant, and these are typically 50,000 to 200,000 square meters, there are multiple handover points between major production steps. Pre-assembly completes and hands off to main assembly. Main assembly completes and moves to test. Test completes and moves to FAT preparation.

At each of those handovers, the ERP should log the transition. In practice, it logs it when someone remembers to scan or enter the update manually. That gap, between when the physical handover happened and when the system knows about it, is where production planners lose the ability to make accurate predictions.

Norman, why does manual scanning fail at this scale?

Norman Dziengel: The honest answer is that manual scanning was designed for a different problem. It works well when you have a discrete, predictable production line where every unit passes through a fixed checkpoint. Energy equipment manufacturing does not work that way.

An order might go through pre-assembly, then rework, then back to main assembly, then to a hold area, then to test. That routing is not always predictable at order-entry time. It evolves as the build progresses. Asking technicians to manually scan at every transition means interrupting highly skilled, highly paid people to do data entry. It does not happen consistently. And when it does not happen, the ERP does not know where the unit is.

"Every hour a finished sub-assembly sits in the wrong staging zone is an hour that your production planner cannot see, and an hour your customer will eventually feel."

— Dr.Norman Dziengel, Head of Product, Inpixon

Thomas Miller: From a commercial perspective, this creates a problem that does not show up in any KPI until it is already too late. A customer calls asking for a status update, which happens constantly on high-value orders. The planner checks the ERP. The ERP says the unit is in final assembly. The unit is actually sitting in a hold area waiting for a missing component. The planner gives the customer the ERP answer in good faith. The customer builds their own schedule around that answer. Three weeks later the unit ships late, and the customer's project is disrupted. That conversation does not stay between the planner and the customer. It goes to procurement. It goes to the category manager who runs the next supplier evaluation. I have seen relationships of five or ten years start to come apart over exactly this sequence of events.

What RTLS delivers at a glance

• Automatic zone-level and sub-meter WIP tracking across the full production footprint
• Real-time ERP and MES posting without manual scanning
• Dwell time monitoring to surface invisible queues and staging delays
• Geofence-triggered process adherence for sequence and timing compliance
• Rule-based process automation based on location data
• Automated traceability records for FAT and customer documentation
• Full indoor and outdoor yard coverage in a single platform

What does the floor actually look like when this goes wrong in practice?

Norman Dziengel: Let me give you a concrete picture. In turbine or large generator assembly, you have a main rotor or stator that moves between stations by overhead crane. Between crane moves, it sits. It might sit for two hours. It might sit for two days. Without an automated tracking system, the only way anyone knows where it is and how long it has been there is if a technician or supervisor physically walks the floor or checks a paper traveler. In a facility with 200 to 300 active work orders in various stages, that is not a scalable process.

What you get instead is a daily firefighting routine that most production planners have simply normalized. The planner calls the floor supervisor. The floor supervisor walks the floor or calls a technician. Someone locates the unit. The ERP gets updated manually, if it gets updated at all. This cycle repeats multiple times a day, for every planner, across every active order. Nobody calls it a problem anymore because it feels like the job. But it is not the job. It is the gap between the job and the system.

Thomas Miller: And the commercial consequence is that the production planner cannot confidently give the sales team a ship date. The sales team gives the customer a ship date with a buffer built in to compensate for the uncertainty. The customer pushes back on the lead time.
The manufacturer ends up quoting longer lead times than are actually necessary. In a market where lead time is the primary competitive differentiator right now, that is a real commercial cost. Shorter, more confident delivery windows win orders. That confidence has to come from somewhere, and it cannot come from a system that is three days behind.

Can you give a concrete example of what changed after RTLS was deployed at a manufacturer like this?

Norman Dziengel: The most direct measure was the 50% reduction in SAP posting deviations. That number reflects the gap between what the ERP thought was happening and what was actually happening, and it was cut in half because the system was now posting automatically based on real physical movement rather than waiting for manual input. The 25% improvement in dwell-time stability is equally significant. Dwell time is how long a unit or component stays in a given zone. When dwell time is unstable, it means units are sitting longer than planned in some zones and rushing through others. That instability propagates forward through the schedule. Stabilizing it is a prerequisite for reliable OTD.

Thomas Miller: The 25% reduction in congestion interruptions is the one that I find most directly connected to the delivery promise. Congestion interruptions, where a zone fills up because upstream flow is not coordinated with downstream capacity, are one of the primary causes of unplanned delay in this type of facility. Eliminating one in five of those interruptions is not a small operational improvement. In a plant producing turbines at 15 to 25 million euros per unit, every day of recovered throughput time has a direct revenue value. 

What holds most manufacturers back from making this change?

Thomas Miller: The most common hesitation I hear is integration risk. "We have SAP. We have a legacy MES. We do not want to add complexity to an already complex IT landscape." That is a legitimate concern, and it has to be taken seriously. The answer is not to dismiss it but to demonstrate that a well-designed RTLS deployment does not require ripping out existing systems. It adds a location layer on top of what is already there. SAP still runs the plan. The MES still manages execution. RTLS provides the real-time physical truth that makes both of those systems more accurate.

Norman Dziengel: There is also a technology question that I think gets overcomplicated. Facilities often assume they need to choose a single radio technology, UWB or BLE or something else, and commit to it everywhere. In practice, different zones have different requirements. Sub-meter precision matters in final assembly and FAT staging. Zone-level tracking is sufficient in the yard or in receiving. A good platform handles multiple technologies under a single data layer, so the operational team sees one consistent picture regardless of what radio technology is running in any given zone.

Where do you recommend manufacturers start?

Norman Dziengel: Start with the handover points. The transitions between major production stages, receiving to pre-assembly, pre-assembly to main assembly, main assembly to test, test to final staging, are where the most significant tracking gaps occur and where the ERP-to-floor divergence is largest. Instrumenting those transitions first gives you the fastest improvement in ERP data quality and the fastest improvement in production planner confidence. You do not need to track every component everywhere on day one.

Thomas Miller: From a commercial standpoint, I would add: start where the OTD risk is highest. If you have a specific product family or a specific customer where late delivery has the most serious contractual or relationship consequences, that is your pilot scope. Demonstrating that a targeted RTLS deployment reduces late deliveries in that segment builds the internal case for broader rollout more convincingly than any general proof-of-concept.

Looking forward, what does this capability enable that manufacturers cannot do today?

Thomas Miller: Shorter, more confident delivery windows. Right now, most manufacturers are quoting with buffer because they do not have confidence in their floor. In a market where transformer lead times average two and a half years and gas turbine lead times are stretching toward five years or more, the manufacturer who can credibly commit to a shorter, tighter delivery schedule has a real commercial advantage. That confidence has to come from the floor. It cannot be manufactured in a spreadsheet.

Norman Dziengel: The longer-term capability is predictive intervention. Once you have continuous real-time data from the floor, where every unit is, how long it has been there, how that compares to plan, you can start detecting delay patterns before they reach the customer. A unit that is trending two days behind plan in pre-assembly is not yet a late delivery. It is still a recoverable situation. But you can only recover it if you see it. Right now, most facilities see it when it is already too late to change the outcome. That is what we are working to change.

If your production planner cannot tell you within ten minutes where any active work order is on your floor right now, the delivery promise you made to your customer is based on a plan, not on reality. See how a leading turbine manufacturer closed that gap, and what it meant for their production performance, in our success story.

To understand how RTLS maps to your specific production environment, visit our production tracking solution page.