Why most maintenance-prioritization models die in a spreadsheet

The graveyard of good models

Every asset-intensive operator has seen it: a consultant or an internal analyst builds a risk model, presents a compelling ranked list, everyone nods — and eighteen months later the budget is being allocated the way it always was. The model was not wrong. It was unmaintainable.

The failure pattern is consistent. The model lived in a workbook only its author understood. Its input data was assembled by hand for the presentation and never refreshed. Its assumptions were defensible but undocumented, so the first hard question in a budget meeting went unanswered. None of these are modelling failures; they are software failures.

What the surviving models have in common

Models that keep informing decisions are delivered as systems. Their inputs arrive through pipelines from the asset register and inspection records, not through copy-paste. Their assumptions are parameters with names, units, and documented sources, visible to the people who challenge the outputs. Their results appear in an interface a planner can use during the meeting, not a PDF prepared before it.

Just as important, surviving models are honest about uncertainty. A ranking that presents false precision gets discredited by its first wrong call. A model that shows confidence ranges — and improves as data quality improves — earns the more durable kind of trust: it is treated as evidence, weighed alongside engineering judgment rather than replacing it.

• Automated data feeds, so the model reflects this cycle's reality
• Named, documented assumptions that stakeholders can interrogate
• An interface planners use directly, not a deck about the model
• Explicit uncertainty, so trust survives the first surprise

Budget the software, not just the analysis

The practical implication for anyone commissioning modelling work: budget for the delivery system, not only the analysis. A model with pipelines, an interface, documentation, and an owner costs more than a workbook — and it is the difference between influencing one budget cycle and influencing all of them.