How long will it take?

Let's start with the mistake that's underneath most of them.

Effort is not duration

Two words that get used as if they're the same thing. They aren't, and confusing them is the most common estimating error there is.

Effort is labour. Forty man-hours. It's how much work exists.

Duration is calendar time. How long it takes, with the crew you actually have, on the site you actually have.

Forty man-hours does not mean five days. It might mean eight.

Nobody works every minute of the shift. There's the briefing, the setup, moving materials, waiting on the crane. Even a good crew turns maybe three-quarters of the day into installed work. That's not laziness — that's construction.

And then availability. Is that crew yours all week, or are they on two other work fronts? A crew you have half the time takes twice as long.

Skip both corrections and you get the optimism illusion: an estimate that would be right if the world worked the way the spreadsheet does. It doesn't. It never has.

Twice the crew is not half the time

Here's the follow-up someone senior always asks: "Fine — so put twice the men on it."

It doesn't work like that, and everyone who's been on site knows why.

The new crew has to learn the job. Somebody has to supervise them. Two gangs on one pour deck get in each other's way. The crane can only lift for one of them at a time. You add people, and you get less out of each one.

And some things don't care at all. Concrete cures in seven days. Put a hundred men around it and it still cures in seven days. Effort-driven work responds to crew. Fixed-duration work laughs at you.

Knowing which is which is most of the skill.

Three ways to get a number

So where does an honest duration actually come from?

By comparison. The last tower like this took fourteen months, so this one will take about fourteen months. Fast, and fine at tender stage — as long as the projects really are alike. Same ground, same height, same country, same client. Change any of those and the comparison quietly stops being true.

By rate. This is the one construction runs on. Your gang lays 40 m² of blockwork a day, and there's 800 m² here, so that's 20 days. It scales, it's defensible, and you can argue it with a quantity surveyor.

But it stands entirely on one thing: is the rate real? Is it from your last three projects, or from a table someone typed in 2009? Your own historical production data is the most valuable estimating asset your company owns, and most companies throw it away at handover.

From the bottom up. Price every work package, then add them up. Slow, expensive, and by far the most accurate. This is what a baseline deserves.

And the rule that governs all three: never trust a single number from a single source. The foreman's estimate, checked against your production rates, checked against what the last project actually did. Where they agree, you're probably right. Where they don't, you've found something worth knowing before it costs you.

The date you promise

Some work is genuinely uncertain. Ground conditions. A method nobody here has used before.

For those, one number is a lie with a decimal point. So take three: the best case, the likely case, the bad case. Weight them — the likely one counts most — and you get an average.

Now the part almost everybody gets wrong.

That average is a coin flip. Half the time the work takes longer than the average — that's what an average is.

So if you promise the client the average date, you have promised a date you will miss half the time. Not through bad luck. Through arithmetic.

The date you commit to sits further out on the curve — far enough that you'll hit it around 85% of the time. That's not padding. That's the difference between an estimate and a commitment, and knowing which one you're being asked for is a professional skill.

Hidden padding is not contingency

Which brings us to the quiet habit that ruins planners' credibility.

Everyone knows the game. The foreman adds a bit "to be safe." The planner adds a bit on top. The PM adds a bit more. Nobody says so. Now there's three layers of invisible fat in the programme, and no one — including you — knows what the real duration is anymore.

And the first time a client's planner finds it, every number you produce for the rest of the project is suspect.

The professional version is boring and it works: estimate honestly, then hold contingency openly. A visible reserve, owned by the project, released by a decision — not smuggled inside forty activities where nobody can manage it.

Hidden padding gets spent by accident. Declared contingency gets spent on purpose.

An estimate of "18 to 22 days, and here's why" is worth more than "20.4 days" pulled from nowhere. The range tells the truth about what you know. The decimal point only pretends to.

Practical insight

Next time someone asks you for a duration in a corridor, ask three things before you answer.

How much work is there, and what's our real production rate for it? Do we actually have that crew, all of it, all week? And is this an estimate or a commitment — because those are two different numbers?

Then give a range and say what it depends on. It takes ten seconds longer and it's the difference between a professional and a hostage.

Key takeaways

✔ Effort is labour hours; duration is calendar time with the crew you really have.
✔ Correct for productivity and availability — nobody installs work for a full shift.
✔ Doubling the crew doesn't halve the duration; some work is fixed no matter what.
✔ Comparison is for tenders, rates carry construction, bottom-up earns the baseline.
✔ A production rate is only as good as the project it came from — keep your own history.
✔ Cross-check every estimate against a second source.
✔ The average date is a coin flip; a commitment sits further along the curve.
✔ Estimate honestly and hold contingency openly — hidden padding destroys credibility.

What's coming next

We have a network. We have durations. The model can finally do arithmetic.

Next week we do it by hand: the forward pass, the backward pass, and the float that falls out of them. It's the one calculation every planner should be able to run on the back of a drawing — because sooner or later the software will hand you a critical path you don't believe, and you'll need to know whether it's wrong or you are.

Time to find the critical path ourselves.