Three clean piles are worth $9,838.

A trigger is something you can see

Week 4 gave every risk a closing date — the moment it stops being possible. The trigger is the other half: the observation that tells you it is happening now, while there is still time to do something.

Most registers get this wrong in the same way. The trigger column contains intentions rather than observations. Monitor ground conditions. Chase the client. Keep an eye on it.

None of those can ever be true or false, so nobody can ever be late.

A real trigger is a thing a person on site could observe on a Tuesday morning without asking anybody's opinion. Refusal recorded above eleven metres on any bore. No signed wayleave in hand by the third of February. A rebar drawing still at revision B eight weeks before its pour. The standpipe reading above 9.4 metres on two consecutive Mondays.

The test is simple. If two people could disagree about whether it has happened, it is not a trigger.

And a trigger needs somewhere to go. The person who observes it is almost never the person who owns the response — the ganger who watches the auger stop is not the engineer who serves the notice. If nobody has told him that a bore refusing above eleven metres is worth telling somebody about, he will simply write it in the daily record and carry on, and the register will find out in six weeks.

The review that closes lines

Then there is the monthly review, which on most projects is a meeting where the same twenty-two lines are read out in the same order and everybody says the same thing they said last month.

A review that is working does three things, and only one of them involves adding anything.

It closes lines whose window has passed. The steel escalation risk died the day the order was signed at a fixed price. It should come off, with a date and a reason, and the money behind it should be released. Week 4 said most registers only ever grow because nobody wrote down the condition under which a line comes off. This is where that bill arrives.

It revises the lines still open, against evidence rather than mood. Which is the rest of this article.

It adds what has genuinely appeared — including the secondary risks created by last month's responses, which Week 13 warned about and which nobody ever writes down.

A register that grows every month is not being managed. It is being accumulated.

Three clean piles

Now the part that almost nobody does properly, and it is the difference between a register that reacts and one that learns.

Piling starts. The first three bores go down and all three reach toe level without meeting anything. What should happen to the rock number?

The instinct is either to do nothing, because three is a small sample, or to declare the problem solved, because three is three. Both are wrong, and there is an exact answer sitting between them.

Set it up the way we have been carrying it. Either the rock extends into the north, in which case around forty percent of the piles are affected, or it is confined to the southern strip, in which case around fourteen percent are. Before drilling, we put the first case at 55%, which gives an expected 11.89 affected piles and $33,875 of exposure at $2,850 each.

Three clean bores are more likely in the second world than the first. Not impossible in either — just more likely in one. Weight the two accordingly and the probability of extensive rock falls from 55% to 29.3%.

Run that back through the remaining thirty-nine piles and the expected number affected drops to 8.43. The exposure is now $24,038.

Three holes in the ground, and $9,838 comes off the register.

Nothing was built cheaper. Nobody negotiated anything. Three bores came up clean and somebody did the arithmetic rather than waiting to see.

Bad news travels further

Now run it the other way, because the shape of the answer is not what people expect.

Suppose those first three bores all hit rock instead. The probability of extensive rock does not rise from 55% to something like 70%. It goes to 96.6%.

Three hits in a row is quite likely if forty percent of the site is affected and very unlikely if only fourteen percent is. So it is powerfully diagnostic, and it settles the question almost on its own. The exposure goes from $33,875 to $52,031.

Put the two side by side. The same three observations, in opposite directions. Good news takes $9,838 off. Bad news puts $18,155 on. Bad news moves the number 1.85 times as far.

That is not pessimism, it is arithmetic, and it has a practical consequence worth carrying around. An early bad week should worry you considerably more than an early good week should reassure you. Rare events are informative when they happen and only mildly informative when they don't.

Releasing money is a decision too

One more thing, and it is the reason most of this never happens.

Revising a risk upward is easy. Something went wrong, everybody saw it, the number goes up and nobody argues.

Revising downward requires somebody to stand up in a meeting and say the project needs less money than it did last month. That is a harder sentence to say than it sounds, because the person saying it gets no credit if they are right and gets asked a great many questions if they are wrong. So the safe move is to leave the number where it is, and on most projects that is exactly what happens.

The consequence is a register that only ever ratchets up, contingency that is never released, and a project that finishes with money it could have used and a project manager who was told there was none.

Cost & Cash Week 5 set the drawdown rule: contingency is released against named risks. Week 4 of this track gave every risk a closing date. This is where those two finally do something — because a risk that closes releases money, and money released in month four can pay for something useful, while money released at handover is just an accounting entry.

Practical insight

Take your register into the next review and do one thing before anything else. Ask, for every line: what have we learned about this since last month?

Not what has happened — what have we learned. Those are different, and the second question finds far more.

Then, for the three largest lines, write down what evidence would make you halve the number and what evidence would make you double it. Write both, before either arrives. It takes ten minutes and it does two things: it stops the number moving on mood, and it turns out to be the fastest way to discover that the trigger column is full of intentions.

And when the evidence is good, say so out loud, in the meeting, with a number. Somebody has to be the person who releases money, and on most projects nobody ever is.

Key takeaways

✔ A register is true on the day it is written and slightly less true every week afterwards.
✔ A trigger is an observation, not an intention. If two people could disagree about whether it has happened, it is not a trigger.
✔ The person who sees the trigger is rarely the person who owns the response. Somebody has to be told what is worth reporting.
✔ A working review closes lines, revises lines and only then adds lines. A register that grows every month is being accumulated, not managed.
✔ Three clean bores take the probability of extensive rock from 55% to 29% and $9,838 off the exposure.
✔ Three bores hitting rock take it to 97% and put $18,155 on. Bad news moves the number 1.85 times as far.
✔ Revising upward is easy and revising downward is career risk, which is why contingency ratchets and never releases.

What's coming next

Sixteen weeks, and every single line on this register has been something that could go wrong.

That is not what the word risk means, and it is not what the standards say either. Uncertainty runs in both directions. The steel market can fall as easily as it can rise. The gang can beat the rate. The client can want something changed in a way that is worth having. The ground can be better than the report.

We have spent four months building a machine that is structurally incapable of noticing any of it, and so has every project either of us has ever worked on.

Next week: the half of risk management that no construction project does, why the upside is systematically invisible, and what it is worth on this job.