What variance measures
Variance is the difference between what you said would happen and what did. It can be applied to any figure in a marketing plan — spend, enquiries, revenue, cost per lead — and it is normally shown two ways at once: the raw gap in units, and the same gap expressed relative to the plan so that large and small lines can be compared on one page.
A variance is described as favourable when the result is better than plan and adverse when it is worse. Note that better does not always mean higher: spending less than budgeted is favourable on the cost line, but if the leads did not arrive either, the pair of variances together tells a very different story than the spend line alone.
One thing to keep straight. In statistics, variance means something else entirely — how widely a set of values is spread around its average. Both meanings turn up in marketing conversations, so say which one you mean when the audience is mixed.
Why variance matters
A report of actuals has no verdict in it. Knowing that a campaign produced a certain number of enquiries at a certain cost is only interesting next to what you expected, and variance is what supplies that comparison. It is the difference between a report that describes and a report that decides.
It also forces a plan to exist. You cannot report variance without having committed to a number in advance, and that commitment is where most of the value sits: it makes assumptions visible before the money is spent rather than after.
Where variance goes wrong
Revising the plan to match the result is the worst habit. Once the target moves every month, every period is on track, nothing is ever learned, and the report becomes decoration. Keep the original plan visible even when it was wrong, and record the revision as a separate line.
Reacting to noise is the next one. Small movements in small numbers are mostly randomness, and chasing them produces constant changes that make the account harder to read. Agree a threshold before the period starts and only investigate variances that cross it.
The third is reporting the number without the cause. A gap on a slide with nobody attached to it will be explained differently by everyone in the room. Every variance worth reporting needs a sentence saying what happened and who is acting on it.
How to act on it
Split each gap into volume and rate before you explain it. Fewer leads than planned because traffic fell is a different problem from fewer leads at the same traffic because the page converted worse, and the two lead to different fixes. Reporting only the combined miss hides which one it was.
Then keep spend variance and outcome variance side by side. Underspend that came from budget pacing problems is a missed opportunity dressed as a saving. Feed the pattern back into the next forecast, and build the comparison into your standing reporting setup so plan and actual appear together by default rather than being assembled by hand each month.
