Half of this curve has no owner.

Why the upside is invisible

It is not that people are pessimistic. It is that every mechanism we built points one way.

The reading list from Week 2 was six documents and every one of them was a record of trouble: the query log, the drawing register, the ground report's bad news, the subcontract enquiries with their gaps. Nobody keeps a log of things that went unexpectedly well, so there was nothing to read.

The workshop question was what could go wrong. Ask a room of experienced people that and you will get twenty-two answers, none of which will be an opportunity, because you did not ask for one.

The heat map from Week 6 has no negative axis. There is nowhere on that grid to put a thing that saves money.

And then there is the quiet one. If the gang beats the rate on the raft, the saving disappears into the measured work and nobody ever knows it existed. The estimator's allowance absorbed it. Upside on a construction project is invisible by accounting as well as by attention.

Underneath all of that is the incentive. Nobody has ever been asked to explain a missed opportunity. Everybody has been asked to explain an overrun.

Three of them are already in your numbers

So put together an opportunity register for this job and see what happens. Five lines, priced the same way we priced the threats.

The gang beating the rate on the raft, worth $18,000 at 35%. The client wanting the substation slab enlarged, $16,000 at 40%. The ground being better than the report, $14,500 at 30%. The steel index falling, $12,000 at 30%. A dry winter, $9,000 at 25%.

Total expected value: $22,900. Against $84,730 of threat, that is not a rounding error.

Now stop, because three of those five are already in the model.

Week 9 gave measured work a range from $688,700 to $773,900. The bottom of that range is the gang beating the rate. Escalation ran from $19,085 to $50,893, and the bottom of that is the index falling. Winter shutdown was ranged from zero, and zero is the dry winter.

Write those three on a separate register and add them to the answer and you have committed Week 11's error in reverse. There we found escalation priced in the estimate and again in the register, and the model paid for market movement twice. This is the same mistake wearing a cheerful expression.

Which leaves $10,750 of upside that is genuinely absent from everything we have built — and a much more interesting question about the $12,150 that is present and has never been looked at.

What the model already knows

Go back to the cost curve from Week 11 and read the left half of it.

The median is $946,796. The tenth percentile is $875,300. Between those two sits $71,496, and it is not a hypothetical. It is our own distribution, built from ranges the project team supplied, describing outcomes this job will actually have one time in two.

Land anywhere on the good side of the median and the average saving against it is $44,514. Land in the best tenth and you are $90,707 better off than the median.

There is not one line anywhere on this project that says what would be done with that money, who would notice it had arrived, or who would be responsible for making it more likely.

Every threat on this register has an owner, a price and a date. The entire left half of the curve has none of the three.

The moves are the same. The window is not.

The techniques are not new and there is nothing to learn. They are the four responses from Week 13 read backwards.

Where you would avoid a threat you make an opportunity certain — get the variation for the enlarged slab signed before the design freezes, and it stops being a probability. Where you would reduce, you enhance: pre-agree a rate for enlarged slabs now, so that if the client asks, the saving is bigger. Where you would transfer, you share — a subcontractor who finds a faster method will only tell you about it if some of the saving is his. And where you would accept, you take it if it arrives, which is fine, provided somebody has written down who books it.

What is genuinely different is the window, and this is the practical heart of it.

A threat gives you time. The rock is there in March and it is still there in June, and a response in June is late but not worthless. An opportunity does not wait. The client will decide about the substation slab in the next six weeks and then he will decide it with somebody else. The gang that beat the rate on the raft moves to another job in April and the method leaves with them.

Threat management tolerates being slow. Opportunity management does not. Which is unfortunate, because the review meeting from last week is built entirely around the first one.

And the release problem again

There is a reason the enlarged slab has never been discussed, and it is the same reason from last week wearing different clothes.

Capturing an opportunity almost always costs something up front. Pre-agreeing a rate takes a commercial manager two days. Sharing a saving with a subcontractor means giving away money you could have kept. Investigating whether the ground is better costs exactly what investigating whether it is worse costs — the same $6,800, the same three holes.

Spending money to make a good thing more likely is a much harder authorisation to get than spending money to make a bad thing less likely, even when the arithmetic is identical. Nobody is ever criticised for failing to chase an upside, because there is no record that it was ever available.

That is the whole problem in one sentence, and the fix is a column, not a philosophy.

Practical insight

Run one workshop and ask exactly one question, phrased carefully: what could happen on this job that would make it finish early or cost less than the estimate?

Not can you think of any opportunities, which produces silence. The specific question produces answers, because everybody in the room has seen a job go well and knows why it did.

Then sort what you get into two piles. Things already inside your estimate ranges — those do not go on a register, they go into a conversation about which end of the range you are managing towards. Things genuinely outside them — those get a line, a value, an owner and a date, exactly like a threat.

And put the date first, because it is nearly always sooner than you think.

Key takeaways

✔ The upside is invisible by construction: the reading list, the workshop question, the heat map and the cost codes all point one way.
✔ If the gang beats the rate, the saving disappears into the measured work and nobody records that it happened.
✔ Five opportunities on this job are worth $22,900 — but $12,150 is the left tail of ranges already in the model.
✔ Listing those three separately would double count the upside exactly as escalation once double counted the downside.
✔ $71,496 sits between P10 and P50 of our own curve, with no owner, no price and no date against any of it.
✔ The four responses mirror cleanly: make it certain, make it bigger, share it, or take it and know who books it.
✔ A threat tolerates a late response. An opportunity closes, and the review meeting is built entirely around the wrong one.

What's coming next

That is the material. Seventeen weeks of it, from a five-line register that came from nowhere to a curve with both halves and a contract that says whose each half is.

Next week is the last one, and it is not a summary. It is about the thing all of this was actually for — which is not the register, the simulation or the number.

Track 1 taught you to see a delay before it happened. Track 2 taught you to see an overrun before it landed. This track was always about the sentence that comes after both: knowing something is coming is worth nothing at all unless somebody acts on it while it is still cheap, and the reason they usually don't has very little to do with arithmetic.