Every organization defers maintenance. That is not a failure, it is a consequence of finite budgets meeting infinite possible work. The failure is deferring without pricing it, which turns a financing decision into an administrative one and removes it from the view of the people who should be making it.
The mechanism
When maintenance is postponed, the cost does not disappear. It moves, and it grows on the way. It grows for reasons that are physical rather than financial: a failure that would have been a component replacement becomes a system replacement; water that would have been a sealant repair becomes a structural remediation; equipment run past its service interval reaches the end of its life earlier, so the replacement arrives sooner than the capital plan assumed.
Put in financial terms, deferral is borrowing. You are declining to spend now in exchange for spending more later. That is a loan. The distinguishing feature of this particular loan is that nobody states the rate, nobody records the liability, and the repayment date is set by physical processes rather than by agreement.
An organization that would require board approval to take on debt of a certain size will routinely accumulate a deferred maintenance backlog several times larger without any equivalent process, because the backlog does not appear on any statement.
Why the framing matters
Because it changes who the argument is with and what the argument is about.
Presented as a maintenance request, the facilities group is asking for money in competition with every other group asking for money, and it is asking on grounds that sound like prudence. Prudence loses to growth in most budget processes, and should, at least sometimes.
Presented as a financing question, the conversation is different: here is the liability, here is the rate at which it compounds, here is what it costs to service it versus retire it. That is a conversation finance is equipped to have and, in our experience, is often willing to have. The obstacle has usually not been unwillingness. It has been that nobody presented the information in a form that made the tradeoff visible.
The practical version: for each significant deferred item, state the cost to address it now, the estimated cost if addressed in three years, and the probability and consequence of failure in the interim. Three columns. That is usually enough to move the discussion.
The estimates will be uncertain, and that is fine
The most common objection to doing this is that the future cost cannot be known precisely. True, and not a reason to skip it. The alternative to an uncertain estimate is an implicit estimate of zero, which is the least accurate number available.
State ranges. Say what you are confident about and what you are not. An assessment that distinguishes between a roof with documented remaining life and a distribution system whose condition is genuinely unknown is more useful than one that assigns confident numbers to both, and it is more likely to survive scrutiny.
Not everything should be addressed
This is where the loan framing earns its keep. If deferral is borrowing, then some borrowing is correct. A building scheduled for disposition in four years should not receive a twenty year roof. An asset serving a function the organization is exiting should be run to failure deliberately, with the decision recorded so that a successor does not mistake it for neglect.
The goal is not to eliminate deferred maintenance. It is to make each instance a decision rather than a default, with the reasoning written down. A backlog that has been reviewed and consciously accepted is a managed position. The same backlog accumulated by inattention is an unmanaged one, and they look identical on a spreadsheet.
Getting it funded
Two things consistently help. First, tie the request to consequence rather than to condition: decision makers respond to what happens if this fails far more than to a condition rating. Second, present it at portfolio level with a ranking, so that partial funding produces a sensible outcome. A request that only works if fully funded tends to receive nothing, whereas a ranked list that degrades gracefully tends to receive something, and something compounds.
This is as much a case building exercise as a technical one, and organizations that are good at it treat it that way. The failure it avoids is not specific to buildings. A ranked list handed to a busy owner dies the same way in any field, and the same argument has been written about technical audits rather than physical ones, in an account of why the findings sit in a backlog for a year: a long unranked list, no effort estimate beside any item, and the whole thing presented as one approve or reject decision, which reliably receives nothing. The remedy is identical. Name the liability, price the alternatives, rank the response, cap what you are asking for this cycle, and let the decision maker choose a level rather than answering yes or no to a single number.