Impending backlog, and why nobody funds it
By webmaster · · 7 min read
Share this article
The argument you lost with your finance director has already been
settled in your favour, in writing, by the National Audit Office. In
Maintaining public service facilities, published on 22 January
2025:
Funding constraints can prevent departments from focusing on
preventative maintenance, which typically delivers the best long-term
value for money, as they have limited funds available once they have
carried out reactive repairs and works required to continue delivering
business operations and comply with statutory duties.
Read the last clause again. It is the half that gets dropped. The
lifecycle plan is not competing with emergencies alone. It is competing
with emergencies plus everything you are legally obliged to do, and it
is third in that queue every year.
So the finding is not in dispute. The evidence you brought to defend
it was.
Your
evidence came from a 2004 American industrial guide
The standard figure, repeated across FM writing, is that reactive
work costs three to four times as much as planned. The citation trail
behind it, where anyone leaves one at all, does not lead to a UK study.
It points at the United States Department of Energy.
So read the document it points at. It is the O&M Best
Practices Guide, PNNL-14788, Release 2.0, published in July 2004 by
the Pacific Northwest National Laboratory for the US Department of
Energy. It lists, among the advantages of preventive maintenance, an
“estimated 12% to 18% cost savings over reactive maintenance program”.
The supporting sentence reads in full: “Studies indicate that this
savings can amount to as much as 12% to 18% on the average.”
Which studies is not stated. The guide names no source for its own
figure, and it credits the US Navy with pioneering the practice, on
ships.
Now the part that should end the argument. The
three-to-four-times multiple is not in that guide. The
most-cited authority for the ratio does not contain the ratio. It
contains a percentage, about a different quantity: 12 to 18% is a saving
on a maintenance programme’s total cost, not the price of one emergency
call-out against one planned visit. Quoting the second and citing the
first is a category error, and it is the category error that has been
circulating for 20 years.
Then there is the silence where the figure should be. The NAO
examined property maintenance across the 17 main government departments
at length. The word “reactive” appears three times in the whole report,
and all three are qualitative. The national audit body looked directly
at this question, concluded that preventative maintenance delivers
better long-term value, and put no multiple on it.
Your finance director was right to distrust the figure. You were
right about the estate. You lost anyway, and you lost on the
evidence.
What
the Cabinet Office will actually say about deferral
There is one costed statement of delay in this territory. It is
narrower than the ratio, and far more defensible. The NAO records it at
paragraph 1.12:
The Cabinet Office has estimated that deferring backlog maintenance
can multiply costs by over 1.5 times over a two- to four-year
period.
Note what this is. It is an estimate, and the NAO does not publish
the Cabinet Office’s method. It is not a claim about reactive
maintenance in general. It is a claim about deferring work you have
already identified, which is the specific thing your finance director is
proposing to do. Used precisely, it survives being checked. Inflated
into the vendor ratio, it does not.
The NHS has a
name for the thing you cannot fund
The strongest material in this argument is 22 years old and rarely
quoted.
A risk-based methodology for establishing and managing
backlog, published by NHS Estates in July 2004 for England, defines
backlog as the cost of bringing assets that have fallen below condition
B back up to it. Table 3.1 ranks condition A to D. Between B and C sits
a category that is neither sound nor backlog: B(C),
“currently as B but will fall below B within five years”.
The footnote to that grade is the lifecycle argument, in government
guidance:
Sub-elements classified as B(C) should not be considered as backlog
until such time as the condition of the sub-element has fallen to below
condition B. Such sub-elements would be expected to be sustained in
condition B by ensuring the required investment is made in sufficient
time to prevent the sub-element falling below condition B.
Chapter 4 gives the category a name, impending
backlog, and instructs you to estimate it over five years and
ensure “funding is available at the right time to expend on the assets
and prevent them falling below condition B, and thereby becoming
backlog”.
Then the cost gradient, which is the part that answers the ratio.
Condition C means major repair or replacement is currently needed, and
major repair is defined in the same table as expenditure “expected to
exceed one-third of the sub-element’s replacement cost”. Condition D is
“operationally unsound and in imminent danger of breakdown”, and
bringing a D back to B is “expected to exceed 50% of its replacement
cost”.
That is not a measured trajectory, and it should not be sold as one.
It is a definitional structure, which is arguably better: the guidance
your backlog return is built on already encodes what an asset costs once
it has fallen, and it gives you both thresholds. A third of replacement
cost. Then half.
The trend, and what it
cost to find out
The NAO’s Figure 6 tracks NHS England’s backlog from £5.6bn in
2014-15 to £13.8bn in 2023-24, in 2023-24 real terms, an average rise of
“11%, or £908 million, every year”. ERIC 2024/25, published by NHS
England on 16 October 2025, puts the cost to eradicate at £15.9bn, up
15.7% on £13.8bn nominally. Around £3.5bn of that has been reported as
high risk, though the risk-band split is worth confirming against the
dataset before anyone quotes it at a board. Across government, the NAO
found a backlog of at least £49bn as of October 2024.
Government officials told the NAO the rise reflects historic
underinvestment, cost increases and aged buildings reaching the end of
their intended operational life around the same time. Not one of those
is an emergency. Every one is what a lifecycle plan exists to see
coming.
What to do
differently, and what it will not fix
The NAO records HM Treasury’s own account of how this is decided:
“the quality of business cases is a determinant in whether departments
secure funding for maintenance”, and those cases “should evidence the
outcomes that planned investment will achieve”. That is not a brush-off.
It is the criterion, stated by the body applying it, and a business case
built on a US federal plant statistic does not meet it. One built on
your own B(C) register, priced, dated by year of fall and risk ranked
against the definition at 5.13 that reserves “high” for what would
“prevent catastrophic failure, major disruption to clinical services or
deficiencies in safety liable to cause serious injury and/or
prosecution”, has a chance.
The honest limits. The methodology is 2004, rests on Estatecode from
2002, and is healthcare guidance for England. Borrowing the A to D
grades onto a university or a council estate is legitimate, but say that
you are borrowing them. ERIC is England only, and the four nations count
this differently enough that their totals are not comparable. And none
of it tells you which of your assets is a B(C), or the year it falls.
Only a survey does that, which is itself the sort of spend that loses to
an emergency.
The argument is not won with a better ratio. It is won by naming the
category that is not backlog yet, and putting a date on it.
