Wastage on site: allowance, loss and arithmetic

Wastage is the difference between the quantity of material purchased and the quantity that ends up in the finished work.

That definition is uncontroversial and almost useless, because the difference has at least five components, they behave completely differently, and nearly every site adds them together into one number and then argues about it.

The five things inside the wastage figure#

Design allowance. Some material is consumed by the nature of the work and was always going to be. Lap lengths in reinforcement. The cover that is thrown away when a bar is cut to length. Mortar joints. This is not waste in any meaningful sense; it is the difference between the theoretical quantity from a drawing and the practical quantity a builder has to buy.

Cutting loss. Offcuts. A standard bar length divided into required lengths leaves a remainder. Good bar bending schedules and good cutting plans reduce it; nothing eliminates it.

Handling and storage loss. Split bags, cement gone hard because it was stacked against a wall in the monsoon, aggregate lost into the mud at the bottom of a stockpile, tiles broken in transit.

Over-ordering and over-issue. Material drawn from the store that was never needed and never returned. This is the largest component on most sites and the least discussed, because it is nobody's fault in particular.

Loss. Material that left the site and was not used on it.

The first two are predictable and should be planned — and priced, because a contractor who leaves them out of his rate pays for them from profit, as quoting a rate you can live with sets out. The third is reducible by ordinary competence. The fourth is a process failure. The fifth is a security matter. Rolling them into one percentage and comparing it to a norm tells you which of those five to work on: none of them.

Norms are a starting point, not a standard#

Every organisation has wastage norms for common materials, usually inherited, usually expressed as a percentage. They are useful as a first check — a figure far outside the norm is worth looking at.

But two cautions.

A norm is an average over conditions that may not be yours. Cutting loss on reinforcement depends on the member sizes and the standard lengths available; a project of small members will genuinely waste more steel than one of long spans, and no amount of discipline changes that.

And a norm quickly becomes a target from below. If the allowance is a certain percentage, consumption tends to rise to meet it, because nobody is questioned while they are inside the allowance. The norm stops being a control and becomes a budget.

Why the number is usually wrong before anybody analyses it#

To calculate wastage you need three quantities: what was purchased, what was issued, and what the work actually required.

The first is generally reliable, because it comes with invoices.

The second is not, because site stores fail on the way out. Receipts are recorded — there is a lorry and a driver and a signature. Issues are recorded later, from memory, or not at all. So the issue figure is systematically low.

The third is the theoretical quantity from the drawings or the measurement book, and it is only as good as the measurement.

Two of your three inputs are soft, and the wastage figure is a difference between them, which means the error in it is the sum of both errors. A number computed from a low issue figure and an approximate theoretical figure is not evidence of anything, and building a control regime on top of it produces arguments rather than savings.

Reconciliation is per-material and per-period#

A single site-wide wastage percentage answers no question anybody has.

The useful unit is one material, over one period, on one project:

• Opening stock
• Plus received
• Less issued
• Equals closing book stock
• Compared with physical stock

Then separately: material issued, against material the measured work required.

Those are two different reconciliations answering two different questions. The first asks whether the store's records are right. The second asks whether the work consumed what it should have. Sites routinely conflate them, and then cannot tell a recording failure from a consumption failure — which are fixed by completely different people doing completely different things.

The materials worth being strict about#

Not all of them. Being equally strict about everything is a way of being strict about nothing.

Rank by value at risk, which is quantity times rate times how easily the material walks. On most projects that puts cement, reinforcement steel and finishes at the top, and it puts a lot of the items people fuss about near the bottom.

For the top three or four materials, it is worth doing the full reconciliation monthly with real counts. For the rest, a quarterly count and a sensible eye is enough. The effort saved on the tail is what pays for doing the head properly.

Measuring loss requires measuring the ordinary#

The uncomfortable conclusion is that you cannot detect theft on a site that does not record its ordinary consumption accurately, because theft is a residual and a residual is only visible when everything else in the equation is known.

This is the same structural point as why measured and docketed quantities disagree. A gap becomes information only when the ordinary causes of that gap have been recorded and subtracted. Otherwise every gap is explicable by any of five stories and the argument goes to whoever is more persistent.

So the sequence is: fix the issue register, fix the units, count the top materials monthly, record differences with reasons — and only then start drawing conclusions from the residual. It is slower than everybody wants and it is the only order that works.

The short version#

Wastage is five different things wearing one number.

Separate design allowance from cutting loss from handling loss from over-issue from actual loss, because each has a different fix and only one of them is a security problem.

And before you interpret the figure at all, check whether your issue records are complete. On most sites they are not, and everything downstream of them is arithmetic on a guess.