Cost Variance
Cost variance is the difference between what work was expected to cost and what it actually cost. In earned value terms it is the value of completed work minus the actual cost of that work: a negative figure means overspend. Reported alone it is ambiguous, because spending less than planned may mean less has been done.
What it contains
A cost variance calculation needs three numbers for a given period: the budgeted cost of the work planned, the budgeted cost of the work actually completed, and the actual cost incurred. The variance is the second minus the third. Comparing only the first and the third — budget against spend — is the version most projects report, and it conflates two different problems: doing less work than planned, and paying more than planned for the work done.
At line level, variance is usually held per cost category: internal labour, vendor fees, licences, infrastructure, travel, contingency drawn. Aggregate variance rarely explains itself; the explanation lives in one or two lines, and a budget workbook that holds cost lines individually is what makes that traceable.
How it is used
Cost variance is reported to answer one question at a governance meeting: is the remaining budget sufficient to finish. That means the variance to date matters less than what it implies about the forecast. A consistent overspend of ten per cent on labour, if the cause is structural rather than one-off, projects forward across all remaining work.
The second use is diagnosis. A variance driven by rate — paying more per day than assumed — is a commercial problem, addressed through the contract or the resourcing mix. A variance driven by effort — more days than estimated for the same output — is an estimating or scope problem. The two look identical in a total and require entirely different responses.
The third is change control. Where the variance traces to approved scope changes, the baseline should have moved with them. The change request log holds the cost impact of each approval, and reconciling it against the budget is what separates real variance from unrecorded change.
Variance figures are usually presented at steering committee with the forecast at completion beside them, since the committee's decision is about future funding rather than past spend. The structure of that report tends to put the number, the cause and the ask on the same page.
Where it goes wrong
The most frequent misreading is treating underspend as good news. Early in a project, underspend almost always reflects delayed mobilisation, unfilled roles or contracts not yet signed. The money will be spent, later and often faster.
The second is variance measured against a stale baseline. Scope has changed three times, the approvals were given, but the budget baseline was never re-issued. Every subsequent variance is then partly real and partly bookkeeping, and nobody can separate them.
The third is missing accruals. Work performed by a vendor but not yet invoiced does not appear in actual cost, so the variance looks favourable until the invoice lands. This is the same defect that distorts burn rate, and it has the same fix.
The fourth is variance reported without a cause. A number in a report with no explanation invites the meeting to invent one. The convention worth holding is that any variance above a stated threshold carries a written cause and a named owner, in the same way governance items do.
Related terms
Burn rate is the speed of spend, independent of work completed. Earned value is the budgeted cost of work actually performed, the middle term in the variance calculation. Schedule variance is the equivalent comparison for time. Forecast at completion projects total cost including the remaining work. Baseline is the approved plan that variance is measured against, and it moves only through change control.
Questions
What is the difference between cost variance and budget variance?
Cost variance compares actual cost to the budgeted cost of the work completed. Budget variance usually compares spend to the period budget regardless of how much work was done.
Is a positive cost variance always good?
No. It can mean efficient delivery, or it can mean less work has been completed than planned. It only reads correctly alongside progress data.
What variance threshold should trigger escalation?
That is set by the governance framework rather than by a general rule. What matters is that the threshold is agreed in advance and applied consistently.
Does cost variance require earned value management?
The formal calculation does. Projects without earned value can still compare cost per completed milestone, which gives a usable approximation.