Rock is not a risk. It's a noun.

You can't manage a noun. You can't put a trigger on one, you can't tell when it has happened, and you certainly can't hold anybody to account for it. Which is roughly why most risk registers are read once and never acted on.

Because, may occur, which would lead to

There's one sentence that fixes this, and it has three parts.

Because [something that is already true], [an uncertain event] may occur, which would lead to [an effect measured in money or days].

Each part does a different job. The cause is a fact — you can point at it, it is already true today, and it is the thing you might be able to change. The event is the uncertain bit, and it has to be specific enough that you could tell somebody the day it happened. The effect is the consequence, in numbers, not in the phrase delay to the project.

Run line one through it:

Because the ground investigation covered five boreholes on a forty-two pile site, rock may be encountered below pile toe level away from the tested locations, which would require a heavier rig, add two days per affected bore and delay the pile cap sequence.

Longer. Much better. And notice what appeared that wasn't there before: an owner is now obvious (whoever controls the ground investigation scope), a trigger is now obvious (the first bore that meets refusal), and a response is now thinkable, because the cause is a thing you could actually do something about.

The test kills eight of them

Put all twenty-two through that sentence and eight of them fall apart. Not because they're unimportant — several of them matter more than the ones that survive. They fall apart because they aren't risks.

Drawings still at revision B has no uncertain event in it. It's a fact, and it's the because half of late detailing. It belongs inside another line, not on its own.

Rebar clash between the core wall and the transfer slab has already happened. Somebody found it. That isn't a risk, it's an issue, and issues get solved this week rather than monitored monthly.

Gang productivity nobody has measured is an assumption. It has no event. You don't assign it a probability — you go and measure it, and if you can't, you write it in the assumptions log where somebody might challenge it.

The neighbour's working-hours restriction is in the planning consent. It is certain, permanent and known. That's a constraint. It belongs in the programme, not the register.

Three of them were never risks. They were money.

This is the group worth stopping on, because it is the most expensive mistake on the list.

Pile trimming. Builder's work in connection with the mechanical package. The existing services diversion.

Every one of those is work that will definitely be done, by somebody, and paid for, by us. There is no uncertainty about whether. The only uncertainty was whose enquiry it should have been in, and that question got settled the moment we noticed — it's ours.

Together they are $30,400.

Now watch what happens if they stay on the register. Somebody gives each of them a probability, because that's what the register asks for. Call it 50%, which is what people write when they don't know. The expected value comes out at $15,200, and $15,200 is what gets carried.

We will spend $30,400.

Writing a certain cost as a 50% risk is a way of only half-paying for it. The other half comes out of the margin, and the margin on this job is $48,163. Three lines in the wrong column and two-thirds of the profit is already spoken for, months before anybody notices.

Those three don't need a probability. They need a number in the estimate and an apology to the estimator.

A register is not an inbox

Here's the pattern underneath all eight.

A risk register collects things that worry people. But worries come in at least five different shapes, and each shape has a different home, a different owner and a different verb.

A risk is managed. An issue is solved. An assumption is validated. A constraint is planned around. A missing cost is priced.

Put all five in one document and every one of them gets the wrong verb. The issue gets monitored instead of fixed. The assumption gets a probability instead of a phone call. The missing cost gets discounted instead of paid.

None of the eight got deleted. Every one of them went somewhere — it just wasn't here.

Every risk has a closing date

One more column, and it's the one almost nobody writes.

Every real risk has a window. It opens on the date the event first becomes possible, and it closes on the date it stops being possible. Steel escalation dies the moment the order is signed at a fixed price. The rock dies when the last bore is drilled. The wayleave dies when the first lorry drives onto the site.

Write both dates and two useful things happen. You know when to stop paying attention, and you know when to release the money you were holding against it — which is the drawdown rule from Cost & Cash Week 5, finally attached to something real.

It also explains something you've probably noticed about every register you've ever inherited. It only ever grows. Nothing is ever taken off, because nobody wrote down the condition under which it would come off.

Practical insight

Open your register and read only the middle column — the thing that is supposed to be the risk.

If a line is a noun phrase with no verb in it — weather, ground conditions, subcontractor performance, design changes — it is not written as a risk, and nobody is going to act on it. Rewrite it with the three-part sentence. It takes about ninety seconds a line.

Then ask two questions of every line that survives. On what date does this stop being possible? And what would I actually see, on site, on the day it happened?

If you can't answer the second one, the event isn't specific enough yet. And if you can't answer the first, that line will still be on the register at handover.

Key takeaways

✔ A risk is a sentence, not a noun: because [cause], [event] may occur, which would lead to [effect].
✔ The cause is already true and is the thing you can change. The event must be specific enough to date. The effect is money or days.
✔ Eight of the twenty-two failed the test — one issue, one assumption, one constraint, two causes folded into other lines, and three missing costs.
✔ A risk is managed, an issue is solved, an assumption is validated, a constraint is planned around, a missing cost is priced. One document gives all five the wrong verb.
✔ $30,400 of certain cost sat on the register. At a made-up 50% it would have carried $15,200 — against a margin of $48,163.
✔ Every risk has a closing date. Steel escalation dies when the order is signed; the rock dies when the last bore is drilled.
✔ Registers only ever grow because nobody writes the condition under which a line comes off.

What's coming next

We started with five lines. We read six documents and got twenty-two. We sorted them into eight branches and found two empty. And this week the twenty-two became fourteen properly written risks, plus eight things that were quietly in the wrong document.

Fourteen sentences, each with a cause, an event, an effect, an owner and a closing date.

Now comes the hard part, and it's the one the whole contingency rests on. Somebody has to put a number against each of them. Next week we go back to the very first line of the very first register and ask the question nobody asked: where did 40% come from? Two of five boreholes hit rock. Is that forty percent? It depends entirely on where those five holes were, and nobody has looked.