Target-actual comparison of cost determinations

Two cost determinations are comparable only where they agree in ordering, scope and presentation. Four attributes must therefore stay constant across the stages: A single break is enough to void the comparison.

Attribute	What must stay constant
Structure	the same leading ordering, normally the cost groups
Scope	the same cost groups are included
Tax presentation	consistently net or consistently gross
Reference quantities	the same area and quantity measurement

If one of these breaks, the difference between two totals is not a cost deviation but a presentational effect. It can be neither managed nor attributed to anyone.

That this is more than a methodological nicety is shown by case law: in proceedings concerning alleged errors in a cost update, it was established that schedules ordered by trade and updates ordered by cost group yield no comparability beyond the total. A submission resting on such documents does not hold.

Four breaks occur regularly, and the last produces the largest apparent effect of all. It rests on no real change in cost whatsoever.

Break	Where it arises	Effect
Switch to procurement packages	from the pre-tender estimate onwards	trades and element groups are not congruent: a drylining contractor works on internal walls, floors and fixtures at once
Switch of depth of treatment	cost estimate at level 2 against cost calculation at level 3	comparison only at the coarser level, by consolidating the finer one
Adding or removing cost groups	e.g. KG 500 carried only later	produces a variance that is not one
Switch between net and gross	tax rate 19 %	apparent effect larger than most real variances

A change of ordering is not always avoidable, since procurement follows trades while design follows elements. What is avoidable is the loss of comparability.

The workable route is to retain the cost group structure as the leading one and carry procurement packages as an additional allocation. Every item then holds two attributes, and analysis is possible in both directions.

Where that does not happen, a reconciliation table allocating each procurement package to its cost groups must be maintained. It is effort, but less than reconstructing it afterwards, and it is the difference between a demonstrable and a contestable cost statement.

An overall deviation is worthless as information. Only attribution by cause makes it usable, both for management and for the question of liability.

Cause	Attribution
Quantity change	determination was right, quantity was not; establish the source
Change of scope	client decision, not a forecasting failure
Changed qualities or standards	a decision, normally recorded
Unforeseen circumstances	typical in existing buildings, points to depth of survey
Market movement and price adjustment	outside the designer's sphere of influence
Divergent unit rates	the real test of the quality of the determination

Only the last row bears directly on the quality of one's own work. Failing to maintain this separation continuously means it cannot be reconstructed later, and in a dispute the whole deviation is then argued rather than the attributable part.

A presentation with four columns per cost group works well, supplemented by two entries that place the position in context. The fourth column is the most important and the most often omitted.

Column	Content
1	allowance from the preceding stage
2	current allowance
3	variance
4	cause of the variance
+	procurement status: what share rests on contracts and what on forecast
+	closed or open: whether a decision has already been taken

This presentation also discharges the duty to inform: where it becomes apparent that the determined costs will be exceeded, the client is to be informed without delay. A comparison kept and handed over continuously is the simplest evidence that the duty was met.

Alongside comparison between the stages sits a second comparison using a different benchmark: comparison against the client's cost target. The two run in parallel and are regularly confused with each other.

The two must be kept apart, because they answer different questions. Comparison between stages asks how the determination developed as design advanced. Comparison against the target asks whether the project stays within the financial frame.

A project can pass the two tests differently. A cost calculation a few per cent above the cost estimate is unremarkable professionally; but if the target sits below the cost estimate, the project was never within the frame.

For presentation it follows that the target is carried as a line of its own, stating the cost groups it covers. A target for cost groups 300 to 500 cannot be compared against a total across all groups, and that error is commoner than the reverse.

Two limits are worth observing so that the comparison is not over-read. Both concern what the figures precisely do not say.

Limit	What follows
It measures agreement, not quality	a determination that does not diverge may simply have been carried over unchanged; a strikingly small variance deserves the same attention as a large one
It does not replace a forecast	comparison looks backwards; for management what counts is which variances are emerging, which is why an update of the items not yet let belongs alongside it, clearly marked as a forecast

Comparison does not occur only at changes of stage. Several years normally separate the cost calculation from the final cost statement, and most deviations arise within that period.

The standard reflects this by making the pre-tender estimate the only repeatable stage and by providing for the tender sum to be prepared several times during the project. Both are therefore instruments of continuous control rather than determinations at a fixed date.

In practice this means a rhythm of comparison driven by events rather than dates: at every award, at every change of scope, at every variation of weight. The system of both stages is covered in pre-tender estimate and tender sum.