We were brought into an eight-plant polymers and chemicals producer to run a data maturity programme ahead of an S/4HANA implementation. What follows is the gap analysis and the outcome model, with the client removed and every figure converted to US dollars.
The five gaps
The asset register looked healthy from the executive dashboards and was not. Five measurable gaps carried almost all of the risk, and one of them was not being counted at all.
What the gaps were costing
Over the reference period the site recorded 81 days of downtime against roughly 61 million dollars of production loss exposure, an implied 757 thousand dollars for every day lost. Slightly over half of those days traced back to equipment failure rather than market, feedstock or planned outage.
Gap findings, in order of consequence
Maintenance plans absent on half the running assets
Coverage at 49% means preventive work was being scheduled from memory and habit rather than from the system. Closing this and BOM completeness together accounts for more than 70% of the recoverable loss in the model.
Bills of material largely unbuilt
At 26% completeness, planners could not reserve parts against a job. Wrench time was lost to store room trips and emergency procurement rather than to maintenance.
Inventory growing on the back of the missing BOMs
This is the gap that shows up in the finance report rather than the maintenance one, and it is the direct downstream consequence of BOM completeness sitting at 26 percent. With no reliable link between an asset and the parts it consumes, stocking decisions get made defensively. Buyers order against memory and worst case rather than against a bill of material, reorder points are set without consumption history to justify them, and nothing has an owner who can prove a part is no longer needed.
The result compounds quietly. Inventory value climbs while availability of the parts actually wanted stays flat, slow and non moving stock accumulates against assets that may already have been modified or retired, and write off exposure builds on a balance sheet nobody is reviewing against the asset register. None of this appeared in the downtime model, because carrying cost and obsolescence are a separate loss pool from production loss.
The chain runs one way. Nothing downstream of the first link can be fixed by inventory policy alone, which is why stock reduction drives that skip the bill of material tend to rebound within two years.
The outcome. Inventory value and slow and dead stock had grown every year across the five years preceding the programme. Nine months after the data maturity work went live, inventory value fell year on year for the first time in five years. Of everything the programme produced, this is the number that reached the balance sheet rather than the maintenance report, and it is the one a finance director will ask about first.
Master data quality below usable threshold
Starting at 66.75%, roughly a third of records could not be trusted for reporting. Rebuilt against an ISO 14224 taxonomy, this reached 95.4% in the first phase.
Failure attribution never established
No sampled root cause review existed linking downtime days to data-addressable causes. Without it, any benefit case rests on an industry assumption rather than the plant's own evidence. We flagged this as the single largest source of uncertainty in the model.
The warehouse is where the strategy shows up
If you want to know whether an asset management strategy is real, do not read the maintenance report. Walk the warehouse. Every weakness upstream lands there eventually, as stock that was bought against a guess, held against an asset nobody re-checked, and written off years later by someone who never knew why it was ordered. The warehouse is the slowest and most honest indicator of asset management performance there is.
On this programme it was also the first place the work showed up in money.
The starting position
Nine out of every ten items in stock were slow moving. The international benchmark sits near fifteen percent. That gap is not a procurement failure, it is what happens when parts are bought and stored without a maintenance strategy telling anyone which asset needs them.
What was built
62,154 bills of material were rebuilt against vendor documentation and operational recommendations, taking BOM coverage of the maintainable asset base to 26 percent. Three spare classes were configured so the structure could carry a real stocking policy rather than a flat parts list.
What it moved
Coverage at 26 percent is not a finished job. It was still enough to break a five year trend. Inventory value fell by 2.4 million dollars, the first reduction in five years, and the first driven by data rather than by a spending freeze. Modelled forward at a conservative two percent volume reduction a year, the same work carries roughly 17.9 million dollars over five years, and materially more if the stocking philosophy is implemented aggressively.
Where the value actually sits
Worth noting what the BOM work pays for. Inventory reduction is the visible headline, but it is the smallest share. Most of the return comes from spares being findable and correct when a job is planned.
What the model says is recoverable
Against the 31 million dollar equipment-failure pool, and using a realisation rate between 25 and 40 percent, the programme models annual loss avoidance of roughly 7.8 to 12.4 million dollars once the remaining gaps are closed.
What is delivered and what is modelled. The move from 66.75% to 95.4% master data quality is a delivered outcome. The loss avoidance range is a projection built on three adjustable inputs: the share of failure days addressable by data quality, the realisation rate, and the ramp. It should be recalibrated against a sampled root cause review of the 41 failure days before anyone commits to it. Neither figure includes inventory carrying cost, obsolescence or write off. Those sit outside the production loss pool entirely and were tracked separately, so the inventory result reported above is additional to the range shown here rather than part of it.
Figures converted to US dollars at the fixed rate applying in the operating jurisdiction. The client, its location and its plant identifiers have been removed.
English (United States)