Brick despatches at Michelmersh fell another 9% in the first half of this year, leaving volumes more than a quarter below the 2022 peak. Revenue came in at £32.4m against £35.8m, the Charnwood site in Leicestershire has closed with 30 jobs lost, and the chief executive described the market as a long trough (Construction Enquirer, 1 September 2026). Put a request for extra kiln capacity in front of that board this morning and every person in the room knows the answer before the paper is opened.
Capital investment decisions almost never arrive that clearly labelled. The numbers are plausible, the payback clears the threshold, the sponsor is credible. And nobody in the room can say what the operation will do differently once the money has gone.
How capital investment decisions actually get made
The Bank of England asked 2,227 firms whether they set an investment hurdle rate, meaning a minimum return a project has to clear before anyone will approve it. Around 70% do not set one at all. Among that 70%, almost 40% replace capital items at fixed intervals and around 20% work to a target payback period. The minority who do use a hurdle rate reported an average of 16.4%, and 30% of firms had not moved theirs in over three years (Bank Underground, 22 August 2024).
One qualifier before that number gets used carelessly. Weighted by how much each firm actually invests, the proportion using a hurdle rate rises to 45%, because the larger spenders are the more likely to have one. A bigger share of the money meets a threshold than the headcount suggests.
Even so, the dominant filter on capital spending in British business is a percentage set some years ago, or a calendar.
Both are respectable. Both are also entirely silent on the operation. A hurdle rate tests whether a forecast return beats a number chosen in a different economic climate. A replacement cycle tests how old the asset is. Neither of them asks what will be different on the factory floor, in the exchange, or on the sites, and that is the question the money is actually being spent to answer.
Business investment rose 1.7% in the second quarter of this year and sits 0.8% above the same quarter in 2025 (ONS, 13 August 2026). Capital is moving again. Where it lands inside each business is a separate matter, and a much less examined one.
The test that takes ten minutes
Goldratt set this out decades ago and it has not been improved on. Three questions, asked in order, on one side of paper.
What will this add to the money coming in. The answer has to arrive as money. Extra capacity and saved minutes are not answers to that question, they are inputs to it.
Then the money going out permanently, counted in whole people, named, or in paid overtime hours. Two extra heads that turn out on inspection to be a quarter of one person and a third of another is a warning about the whole case.
Last, what it ties up to get there. The purchase price is the visible part. Stock, buffers and work in progress move with the decision, and that is the part which usually goes unlisted.
The first question is where nearly every request dies, and it dies for a structural reason rather than a lazy one. The output of an operation is set at its tightest point. Spend money anywhere else and the extra capacity has nowhere to go, so the money coming in does not move. The request can be perfectly sound engineering and still fail on that arithmetic, which applies just as hard to a business with a strong balance sheet as to one without.
Even at the tightest point the gain has a ceiling, because relieving one constraint hands the problem to the next one. Double the capacity of the machine that sets the pace and the market may only absorb a fifth of it. The honest number in the paper is the fifth.
The constraint is rarely the equipment
A workforce report published this morning puts the position in UK telecoms plainly. As many as 200,000 engineers could leave the industry by 2030, against a graduate pipeline of around 14,000 a year. Sixty per cent of engineers in the sector are over 50 and 3% are under 35, and roughly 95% of firms say they are struggling to hire (ISPreview, citing the Digital Connectivity Forum, 2 September 2026).
Any operator reading that can work out what it does to a capital plan. Vans, splicing kits and blown fibre machines are all available on a lead time measured in weeks. The person cleared to work unsupervised on a live network is not.
I built operations for a fibre network during its growth phase, and the binding constraint moved around constantly without ever settling on plant. It sat on the number of people who could be trusted alone on a job, on access to the ground, and on whether the survey had been done properly before a crew was dispatched. A second rig does nothing when one person is signed off to use it. An investment case that opens with an equipment specification has usually decided its own answer before anyone has looked at where the work is actually held up.
That is worth carrying into the room where the operation is being judged on what it can absorb rather than on what it owns.
The exception that proves the test works
An investment away from the tightest point can still be the right call, and the reason has nothing to do with output.
Goldratt’s own worked example is a machine bought for a quarter of a million dollars on a resource that was never the bottleneck. It allowed the buffer in front of the plant to come down from three weeks to two, and released several million in stock across the site. That case is obvious the moment the third question is asked, and invisible to anyone asking only the first.
The rule is that a request has to name which of the three it moves, and by how much. Where it moves the third one hard enough, the first can sit at zero and the answer is still yes.
What happens when you re-run the pending list
Two things tend to happen.
Most of the list comes back. Sponsors who can describe the technical merit of a purchase in detail cannot put a number against what it adds to money coming in, and once that is the first question on the form, the request goes back to be rewritten or withdrawn.
Then a second wave arrives, and it looks nothing like the first. Small items. Unglamorous ones. A tool, a licence, a second set of test equipment, three weeks of a contractor’s time to clear a backlog at the point where everything queues. The returns on that wave are frequently absurd, because the spend is tiny and it lands exactly where the output is set.
There is a credibility cost hiding in the gap between the two waves. The supervisor who cannot get five hundred pounds approved for something that would relieve the pinch point, in a business that has just signed off a machine nobody needed, draws the obvious conclusion about how decisions get made. That conclusion travels. It is one of the quieter reasons why an honest look at how an operation really runs so often starts with the capital file.
A buyer will read the same file eventually. Three years of approved spend, and what changed in the business as a result, is one of the cheapest reads available on how a management team thinks. I would not put a number on what a thin answer costs at the table, and anyone offering one is guessing at it. The direction is consistent enough to plan around.
Four questions worth more than the appraisal form
Where is the constraint, by name. Which team, machine, approval, licence or person sets the pace of the whole operation this quarter. Ask three directors separately, and if the answers differ, the appraisal process has been running blind and the rest of this is academic.
Then the pending list. One sentence per request on what changes in money coming in, and by how much.
Behind that sits the retrospective half, which is harder and more useful. Every request approved in the last two years, checked against the operating number it was supposed to move rather than against its own project report. Where that comparison has never been made, it is worth making once and reading the pattern rather than the individual cases.
And the cheapest of the four. Ask the people working at the constraint what they would spend five thousand pounds on. The answers tend to be specific, tend to be already known to their supervisor, and tend to have been waiting a long time. The pattern across most operations I have looked at is that the good small requests were never turned down. They were never asked for, because the form was built for a different kind of spending.
Capital is moving again across the UK economy. Whether this round produces anything depends on a question that takes ten minutes to ask and that most appraisal processes never reach. What will be different in the operation on the day after the money is spent, and who is willing to put a number against it.
References
- Construction Enquirer. “Brick maker battles ‘long trough’ in construction.” 1 September 2026. Read Article
- Bank Underground (Bank of England). “High hurdles: evidence on corporate investment hurdle rates in the UK.” 22 August 2024. Read Article
- Office for National Statistics. “Business investment in the UK: April to June 2026 provisional results.” 13 August 2026. Read Article
- ISPreview UK. “DCF Estimate 200,000 Engineers Could Leave UK Telecoms Workforce by 2030.” 2 September 2026. Read Article
