SMETA variance analysis — reading the numbers before the project ends
Most SMETA overruns are obvious three weeks before they land. Here's how to read line-by-line variance so the director sees the problem while it is still fixable.
Every project director has had the same conversation. Three weeks before closing, the numbers do not work, and everyone is surprised. The thing is — the numbers almost never started being wrong three weeks before. They started being wrong three months before, and nobody was reading variance line-by-line. A proper variance review is a 45-minute weekly exercise that catches the overrun when it is still a 3% deviation, not a 20% crisis.
The three variance types that matter
Quantity variance. Planned quantity versus actual consumed. A column that was supposed to take 1.8 m³ of concrete took 2.1 m³. Quantity variance is almost always a specification or execution issue — the form was wrong, the pour was sloppy, the spec changed and no one updated the SMETA.
Price variance. Planned unit price versus actual unit price. The cement was budgeted at 580,000 / ton and arrived at 620,000. Price variance is a procurement issue — bad supplier choice, weak negotiation, or genuine market movement.
Scope variance. Planned line item versus actual line items. The electrical work budgeted 200 m of cable; the execution needed 240 m of cable plus junction boxes that were not in the SMETA at all. Scope variance is a design issue — the drawings were incomplete or changed mid-project.
A variance analysis that does not separate these three is useless, because the response to each is completely different. Quantity variance = talk to the prorab. Price variance = talk to the procurement lead. Scope variance = talk to the designer.
The weekly 45-minute ritual
Every Monday morning, the director sits down with the finance person and the SMETA. They look at three things:
The five largest absolute variances this week. Not percentages — absolute sums. A 2% variance on a 200 million soum line is more important than a 40% variance on a 5 million soum line.
Every line that crossed a threshold. We recommend 5% of line budget. If a line was within 5% last week and is at 6% this week, it is flagged — not punished, flagged.
Lines trending the wrong way. A line that moved from 0% to 2% to 4% over three weeks is moving faster than one that jumped straight to 4%. Momentum matters.
The "why" column is mandatory
Every flagged variance gets a one-sentence explanation on the review sheet. Not "to be investigated" — an actual explanation. If the explanation is "don't know yet," that is a red flag in itself. Variance without a reason is variance that will compound.
The explanation gets signed by whoever owns the line — prorab for consumption, procurement for prices, design for scope. Ownership is what turns variance from a report into action.
The three-week rule
If a variance has been flagged for three consecutive weeks without closure, it goes to the top of the director's agenda. Three weeks is enough for anyone to have investigated and either closed the gap or escalated it. Beyond three weeks, the variance is no longer a data point — it is a process failure.
What a clean variance report looks like
At the bottom of the weekly sheet, one summary row: total budget consumed, total budget remaining, weighted average variance across all open lines. A project on track runs a weighted average variance of 0% to +3%. A project in trouble runs +8% or higher. The number tells you whether to keep working or call an emergency meeting, in ten seconds.
What happens when firms do not do this
Two predictable failure modes. First — all the variance lands at month-end close, when it is too late to fix, and the director is forced to either absorb the loss or have a very uncomfortable conversation with the client. Second — the variance is hidden in aggregate reports, because the numbers look fine at the category level even though specific lines are hemorrhaging. Line-by-line is not optional.
The real reason weekly variance matters
Variance that is read weekly creates a feedback loop — prorabs know their consumption is being watched, procurement knows their prices are being watched, designers know their scope is being watched. The watching is not punishment; it is signal. Good teams respond to signal by correcting early. Bad teams get to month-end close with no idea how they got there.
Most project overruns are not catastrophes. They are boring, slow drifts that no one was reading. A 45-minute Monday review is the smallest intervention that catches the drift while it is still small.