How to Estimate Task Durations That Hold on Site

Why do most construction schedules slip from the very first week?

Most slippage starts not on site but in the planning room, in the moment when a contractor assigns a duration to a task without knowing exactly how long it takes. Four days for block work, two weeks for finishing — numbers that feel right but are never calculated. When a duration is a guess, it is almost always optimistic: the planner imagines a full crew, no rain, materials already waiting, and no rework. Real sites deliver none of those conditions reliably. The slip does not happen because the crew is lazy; it happens because the plan was never honest about the pace at which the work can physically be done. Getting durations right before the schedule is published is the single highest-leverage step a contractor can take to finish on time.

The three numbers that determine every task duration

Every task duration comes down to three inputs: total quantity of work, daily output rate per worker, and crew size. Divide the total quantity by the crew's combined daily output and you have a raw duration. For block work: 150 square metres of wall, three masons at 20 square metres each per day, gives 150 divided by 60, which is 2.5 working days. For concrete, it is volume poured per shift against pump capacity. The formula is the same for every trade. Everything else — weather, rework, material delays — goes on top of that base figure as a deliberate buffer, added separately rather than hidden inside the duration estimate.

How to apply the formula to a real task

Take plastering. The quantity is 300 square metres of wall. A plasterer covers roughly 90 square metres per person per day in good conditions. With two plasterers, the crew output is 180 square metres per day, giving 300 divided by 180, which rounds to 1.7 days — call it two days for the raw task. Add twenty percent for corners, window reveals, and the staircase junction, and the honest duration is two and a half days. That entry in the Gantt planning view gives the downstream trade an honest start date. A one-day entry, even if it feels faster, just moves the problem to whoever is waiting for the walls.

A worked example: tiling a 120 sqm house in Nonthaburi

Khun Somchai is tiling 120 square metres of floor tile and 45 square metres of bathroom wall tile, 165 square metres in total. His two tilers cover 30 square metres per day on open floor and 20 square metres per day in the bathrooms where cuts and grouting slow them down. The open floor takes four days. The bathrooms need three, but the sequence is tighter: tile one day, cure overnight, grout the next. He adds one buffer day for the job overall, giving eight working days in the plan. Tile delivery is booked two days before the start. He enters the task as eight days in SiteBoard, links grouting to start after tiling ends in each room, and the Gantt shows accurate knock-on dates for every trade that follows. The job finishes on day eight.

Why do experienced contractors still underestimate?

Even experienced contractors set durations that prove optimistic for three consistent reasons. First, the full-crew assumption: the plan imagines all assigned workers present every day, while actual attendance rates on Thai sites rarely exceed 85 to 90 percent across a multi-week task. Second, the six-day week: many plans assume Saturday works, but crews often stop at noon or take the day entirely. Third, rework time: no estimate includes the hours spent breaking out and re-doing a tile section set on a sloping bed. Track actual versus planned completion on every task and the same gap appears project after project until you start adjusting your rates.

Building a rate library from your own past jobs

The most reliable duration estimate you will ever produce comes from a past project of similar scale and trade mix. Every time a task closes in SiteBoard, the planned and actual dates are on record. After four or five projects you can calculate an average performance ratio per trade — say, block work running at 90 percent of planned pace — and apply that ratio to the next schedule. The library compounds quickly: after a dozen jobs you have calibrated rates for every trade you use regularly. Collecting this data requires no extra work; it only requires reading the completion record your on-site progress updates already generate.

Where should the schedule buffer live?

The wrong approach is to pad every task individually. When every task carries a private day of slack, the crew uses it, and the project takes exactly as long as the pads allow — Parkinson's Law in action. The better approach is a single deliberate buffer at the end of each phase, visible in the plan, sized at ten to fifteen percent of the phase duration. A four-week structure phase gets three to four buffer days at the end, not spread across every task. This keeps individual task owners honest, makes the buffer explicit and measurable, and gives the project manager a clear cushion to protect as the phase runs.

Matching your plan to the crew you will actually have on site

A duration calculated for eight workers is wrong the moment you know only six will show. Before publishing the schedule, confirm headcount with each trade leader for the weeks the work falls in. In Thailand, crew availability shifts with agricultural cycles, public holidays, and competing projects in the area. School-holiday seasons pull migrant workers home; harvest time in the northeast does the same. A schedule that accounts for a fifteen percent crew reduction in mid-April is realistic. One that assumes uniform attendance all year is not. SiteBoard's Gantt planning view lets you adjust task durations before work starts, so the crew size in the plan matches the crew you can actually field.

FAQ: What if I have no production data for a trade I'm using for the first time?

Ask the subcontractor. A competent sub knows their own production rate and will give you a duration if you share the quantity and site conditions. Build in a fifteen percent contingency over whatever they say, since subcontractors also tend toward optimism when they want the contract. If you are pricing an unfamiliar trade, get estimates from at least two subcontractors — their numbers should agree within twenty percent. A large gap signals that one has misread the scope and needs clarification before the task enters the schedule. The first time you use a new trade is also when you watch their daily output most closely and record the result for the next job.

FAQ: How precise do durations need to be for the Gantt to be useful?

You do not need to estimate to the half-day. A construction schedule is not a timetable. Individual duration accuracy of plus or minus twenty percent is sufficient for a working plan. What matters more is the dependency structure — which task blocks which — and the habit of updating actual progress in SiteBoard every day so that slippage is visible the week it happens rather than the month it compounds. A schedule where every task is within twenty percent of reality, with honest dependencies linking the trades, will finish far closer to the planned date than a schedule of precise estimates that nobody updates.