COBOD
Finding the cost of a product that was never the same twice
/images/cobod/hero.webp
A COBOD BOD2 gantry 3D printer mid-build on a round concrete house, outdoors under a blue sky
- Client
- COBOD International
- Context
- Innovation Pilot, DTU
- Year
- 2022
- Role
- Facilitation, stakeholder interviews and analysis
- Team size
- 5
- Tools
- Process optimisation · Business case · Stakeholder interviews · Cost analysis · Manufacturing
- Contribution
- Team of 5, facilitated the process and ran stakeholder interviews
COBOD had sold 41 printers and only two of them were identical, which quietly cost R&D a seventh of its capacity. We put a number on it: 5.3 million DKK over five years, with no capital investment.
The brief, and the problem underneath it
COBOD asked us to look at production planning, since their rapid growth had left production unable to follow demand, and the brief we were given was about optimising flow, resources and communication between departments. However, after two weeks of interviews across R&D, production, service and supply chain, the problem we kept arriving at was not in production at all, and we ended up reframing it as: when R&D develops a new feature for the printer, they often have to retrofit it for customers with older versions of the printer, which is time consuming and inefficient. This is due to the way the company released improvements, since every new idea was implemented into the next printer as soon as it was ready, which over time left COBOD with a product that was never quite the same twice.
What the numbers showed
The clearest statement of the problem came from the supply chain manager, who told us that "of all the 41 printers we have produced there is actually only two of them that are the same", leaving 39 different variations of the BOD2 in the field. Here the cost became visible. By analysing R&D's own work distribution over three months, we found that 1,074 hours had gone into retrofitting alone, and together with COBOD we estimated that between 50 and 75 percent of all time spent on the BOD2 went to retrofits, amounting to an average of 358.59 hours a month, or roughly a seventh of the total capacity of a department whose actual purpose is to develop new machines. An R&D engineer described the mechanism plainly, saying that when an early customer asks for an upgrade, "then we can go in and have a completely separate solution made for them". Furthermore, the cost was compounding rather than stable. Since each printer sold adds another configuration that may later need its own retrofit, we modelled the hours as growing with the installed base, and the question we put to management was what happens when 200 printers need customised upgrades rather than 41.
The solution
We proposed two changes, neither of which required an investment. The first was a fixed 12 month release interval, where design changes accumulate and are implemented once a year, which means production builds one version of the printer per year instead of treating every unit as unique. The second was an upgrade kit released with every new version, which preserves the upgrade path for existing customers, and since a kit now bundles a year of improvements rather than a single feature, it is both easier to produce and more valuable to the customer when installed. It is important to note that this does not solve the problem immediately, since the 39 existing configurations still exist and still have to be served. What it does is stop the spread from growing, and it does so through a change we estimated could be implemented within a few days.
What it is worth
We built the business case on COBOD's own hour data, with three scenarios covering five years of accumulated salary cost in R&D: a best case of 10,561,073 DKK under the current model against 3,520,358 DKK under the proposed one, a saving of 7,040,715 DKK; a most likely case of 7,973,955 DKK against 2,657,985 DKK, a saving of 5,315,970 DKK; and a worst case of 5,776,214 DKK against 1,925,405 DKK, a saving of 3,850,809 DKK. The most likely case recovers roughly 66 percent of the time currently spent on retrofitting, and the saving appears in every scenario we modelled. The part I find most useful about this case is that the cost structure does not change and no investment is required, which means the saving is not a payback calculation but simply time returned to the department, alongside secondary gains in production and service that we did not attempt to quantify.
My role
We were a team of five drawn from four different study lines, with only two of us coming from Process and Innovation, which meant I carried a larger share of the methodological work, explaining how to run an interview effectively and why the methods were worth following in the first place. There was noticeable pushback during the problem phase, since several team members were ready to generate solutions before we had defined anything, and holding the team in the define phase of the double diamond is probably the single most useful thing I did on this project, given that the reframing described above is what the whole case rests on. Furthermore, the company was slow to respond, and rather than waiting I went looking for access elsewhere, which is how we ended up speaking to an operator of a COBOD printer through our own network and restarted the project when it had stalled. I also pushed the team to treat R&D's claim that service had no problems as a secondhand source rather than a fact, and interviewing service directly is what showed the opposite.
Reflection
Given more time, I would have observed R&D rather than relying on interviews, since observation would likely have revealed more retrofit time than people are willing to admit to, and in this case we were fortunate that the hours had already been logged. Eight weeks also seems to be too short for proper validation, and several assumptions in the model, especially the growth rate of the installed base, would need testing before anyone acted on the numbers. At the Demo Day the feedback from COBOD was that they "liked that we pointed out savings", and the sentence I took away was "I think you guys found a crucial problem", which is a better outcome than being right about the problem we were originally handed.
/images/cobod/hours-current.webp
Line chart titled Hours pr month spent on BOD2 retrofitting upgrades, current model climbing to roughly 29,000 hours by month 60
/images/cobod/hours-comparison.webp
The same chart with a second, much flatter line added for the proposed model, reaching roughly 10,000 hours by month 60 against the current model's 29,000
/images/cobod/upgrade-kit-process.webp
Diagram showing printer version V1 leading into an upgrade kit icon, leading into version V2