Estimating Activity Durations the Field Will Believe

A schedule is only as good as its durations, and durations are where most schedules quietly fall apart. The logic can be perfect and the critical path can be clean, but if the superintendent looks at "Hang Drywall Level 3 — 4 days" and laughs, nobody will use the schedule for anything except the monthly pay app. This post covers where durations should come from, how to calculate them from quantities and crews, how to handle uncertainty without padding everything, and how to get the field to sign off.

Where durations come from

There are four honest sources, roughly in order of how much the field will trust them:

1. The people doing the work. The foreman or sub's PM who will run the crew. Best source, as long as you ask the right question.
2. Quantities and production rates. Takeoff quantity divided by what a crew does per day. Defensible and checkable.
3. Your own history. Actual durations from similar past jobs. Good if you kept as-built data. Most people did not.
4. Published or estimating-database rates. Useful as a sanity check. Rarely trusted on its own, because it does not know your site, your crews, or your weather.

And one dishonest source: working backward from the contract finish date. Everybody does this a little. The problem is that durations reverse-engineered from a target look exactly like real ones in the schedule, and nobody can tell which is which six months later.

The basic calculation

For most production work, duration is quantity divided by crew output:

Duration (workdays) = Quantity / (Crew output per day)

Worked example. Drywall hang on Level 3 of an office building:

Quantity (from takeoff):        24,000 SF of board
Crew:                           2 hangers + 1 helper
Crew output assumed:            2,000 SF per crew-day
Number of crews:                2
Duration = 24,000 / (2,000 x 2) = 6 workdays

The production rate here is an illustration, not a standard. Use whatever your drywall sub says their crews actually do on this kind of work: ceiling height, board type, number of layers, and how chopped-up the floor plan is all change it a lot. The point is the arithmetic is visible. If the sub says "we'll have three crews, not two," you can recompute in your head and get 4 days. If someone just typed "6" with no basis, there is nothing to discuss.

Write the basis down. A short note on the activity (quantity, crew, rate) costs ten seconds and saves an argument later.

Not all durations are production

Production math works for repetitive work: concrete placement, framing, drywall, ductwork, roofing, curtain wall panels. A lot of construction activities are not like that. They fall into a few other types, and each needs a different approach.

Submittal review deserves special mention. The contract may say the architect has 10 working days to review. In practice, the duration you should carry is the review period plus your own prep time, plus a resubmittal cycle for anything that is likely to come back "revise and resubmit." If the steel shop drawings drive the steel delivery and the steel drives the job, a single resubmittal cycle is a critical-path event. Show it.

Handling uncertainty without padding

Every scheduler is tempted to add a few days to everything "just in case." The trouble with spreading contingency into individual activities is that nobody can see it, it gets consumed whether it is needed or not (work tends to fill the time it is given), and you cannot tell later which slips were real.

A better approach for activities with real uncertainty is a simple three-point estimate:

• Optimistic (O): everything goes right.
• Most likely (M): what usually happens.
• Pessimistic (P): a bad but realistic case, not a disaster.

The common PERT-style weighted average is:

Expected = (O + 4M + P) / 6

Example for excavation and shoring of a basement in uncertain soils:

O = 15 days, M = 20 days, P = 32 days
Expected = (15 + 80 + 32) / 6 = 127 / 6 = 21.2, call it 21 days

Carry the most likely or the expected value in the activity. If you want contingency, hold it as a visible buffer activity before a key milestone, where everyone can see it and decide whether it gets used. Whether the contract allows that, and who owns it, is a separate question covered in Who Owns the Float?.

Getting the field to buy in

The single biggest improvement you can make to durations is to stop guessing and ask the people who will do the work. A few things make those conversations go better.

Ask about crews and sequence, not just days. "How long will Level 3 drywall take?" gets you a number pulled from the air. "How many hangers will you have, and will you start at the core or the perimeter?" gets you a plan, and the number falls out of it.

Ask what has to be true first. The foreman will tell you "I need the in-wall inspection signed off and the floor clear of the electrician's material." That is logic, and it is often more valuable than the duration.

Show the math back. "So 24,000 SF, two crews, 2,000 a day each, that is 6 days. Does that sound right?" People correct numbers more readily than they generate them.

Do not accept "it depends" as a final answer. It does depend. Write down the assumption it depends on and use it.

Watch for sandbagging and for heroics. Subs sometimes inflate durations to protect themselves. They also sometimes promise durations they cannot hit to win approval. Production math is your check against both.

This is a lot of what a good pull planning session does well. If you run them, How to Run a Pull Planning Session covers getting durations and handoffs straight from the trades.

Durations and calendars

A duration means nothing without a calendar. "10 days" on a five-day workweek calendar is two weeks. On a six-day calendar it is a bit under two weeks. On a seven-day calendar used for concrete curing it is ten calendar days.

Be consistent about units. Most schedules use workdays for construction activities. Cure times and fabrication lead times are often quoted in calendar days, so either put those activities on a seven-day calendar or convert them. Mixing the two silently is how a "28-day cure" ends up taking almost six weeks in the schedule. Calendars, Holidays, and Weather Days goes deeper on this.

How long is too long

Long activities hide problems. If an activity runs 60 days, it can be 30% complete for weeks without anyone noticing it has stalled. Common guidance, including the DCMA 14-Point Assessment, flags activities with very long durations for review. For construction work, a rule of thumb many schedulers use is to break any production activity longer than roughly a month or one update period into pieces by area or floor, so you can actually see progress. Procurement and fabrication activities can reasonably run longer because nothing about them is visible on site anyway.

Very short activities have their own problem: too many of them make the schedule hard to maintain. See How Many Activities Should a Schedule Have? for finding the middle.

A quick review checklist

Before you lock durations for a baseline, scan them with a few questions:

• Does every production activity have a quantity and crew basis written down somewhere?
• Do submittal and approval durations include a realistic resubmittal for critical items?
• Are long-lead fabrication durations backed by a vendor quote, not a guess?
• Are cure and lead times on the right calendar?
• Are any activities longer than one update period that could reasonably be split?
• Did someone from the field look at the durations and agree with them?
• Is contingency visible as its own activity, rather than buried in every duration?

Key takeaways

• Durations should have a visible basis: quantity, crew, and rate, or a vendor/spec source.
• Different activity types need different estimating methods. Not everything is production.
• Use three-point estimates for genuinely uncertain work, and keep contingency visible instead of padding everything.
• Ask the foreman about crews and sequence, then show the math back. That is how you get durations the field believes.
• Match the duration to the calendar it runs on, especially for cure and lead times quoted in calendar days.