Variance analysis: the reason behind plan versus actual

Last working day of the month at a fifty-person industrial supplies company. The bookkeeper circulates the plan-versus-actual sheet. The revenue line sits below target and the cell is red. Three people in a row say the same thing at the review: next month we push more quotes, more visits. Nobody says the other thing sitting in the same sheet — units sold came in above plan. The entire gap came from average selling price eroding, because discount authority in the field had quietly widened. That same meeting repeated three months running, more quotes went out for three months, and the gap never closed. The number was seen every time. The cause was never discussed.

Variance analysis fills exactly that gap: it stops the difference between plan and actual from being one number and breaks it into components, each of which attaches to a different action. What follows covers why a variance is a question rather than a finding, how to decompose the gap into volume, price and mix, which variances are worth explaining at all, how to tell timing from permanence, why favorable variances deserve the same scrutiny, the cases where the plan itself is the defect, how to structure the review meeting, and how to get from an explanation to a decision.

Total varianceComponentsCauseAction

A variance is not a number, it is a question

Most plan-versus-actual sheets produce a color and stop there: red, amber, green. But the variance itself is not a finding; it is a marker telling you where to go looking for one. Revenue is below plan carries exactly as much information as the patient has a fever — it announces that something is off without saying what. Teams that read the sheet and decide immediately are prescribing before diagnosing.

Set the sequence up this way: the difference is an observation, the cause is a hypothesis, the action is a decision. Most teams leap from observation straight to decision and drop the hypothesis in between, which is why the decisions they take have nothing to do with the variance. Issuing more quotes does not close a gap created by price; it only tires the team out. None of this works unless everyone calculates the same metric the same way first, which we covered in our piece on the metric dictionary.

Decomposing the gap: volume, price, mix

There are at most three fundamental components behind a revenue variance, and each attaches to a different person, a different decision and a different timeline. The volume component is about how many units you sold. The price component is about what you received per unit. The mix component is about how much space a given product or segment occupied in the total. A gap of identical size can be produced by any one of the three, and the remedies have nothing in common.

Keep the order fixed when you decompose, or the same difference gets counted twice. A workable convention: calculate the volume difference at plan price first, then the price difference at actual volume, and leave the mix effect as the residual. The figures do not need accounting precision. The point is not to balance a ledger; it is to put the cause in the right box.

• Volume variance: Fewer or more units sold than planned; the cause usually sits at the top of the funnel, in whether enough opportunities were created at all.
• Price variance: Unit price landed away from plan; loosened discount authority, an unapplied list update, or slackening negotiation discipline are the typical sources.
• Mix variance: The right volume sold, but low-contribution products took a larger share; a shift that looks minor in revenue and lands hard on margin.
• Timing variance: The sale did not fail to happen, it fell on the other side of the period boundary; the most common illusion of a permanent loss.
• Scope variance: A product, region or channel that did not exist when the plan was written entered or left the period; it corrupts the baseline and inflates the difference artificially.
• Input and rate variance: Nothing changed in selling conditions, but movement on the cost or exchange-rate side changed the outcome; this is outside the sales team's control and belongs on its own line.

The most practical benefit of this split is that it shortens meetings. One line naming the dominant component turns a half-hour of general discussion into a thirty-second observation. To catch movement on the price and mix side early, track average deal size as a separate line; how to measure it is in our article on average deal size.

Which variances are worth explaining?

A team that sets out to explain every variance soon does nothing except explain variances. You need a threshold, and it has to work in two directions: proportional and absolute. Set only a percentage and large swings on small lines will own the agenda. Set only an absolute figure and small lines deteriorating quickly slip past unnoticed.

The second filter is controllability. If the cause of a variance can be changed by decisions the team makes, it goes on the agenda; if it cannot, it gets recorded and skipped. That distinction also matters for morale: a team asked to account every month for a line nobody controls stops taking the sheet seriously, and the sheet becomes a ceremonial object.

An unexplained variance does not just blur the period behind you; it corrupts the ground the next plan will be built on.

Timing variance or permanent variance?

The most expensive mistake in period-end reporting is counting a deferred deal as a lost one. A contract not signed in the last week of the month but signed in the first week of the next is a shift, not a loss, and it deserves a different response. A lost deal justifies a campaign; a shifted deal does not — it justifies managing the period-end close.

The simplest way to tell them apart is to read the opportunities behind the gap one by one: is the record still open, did the stage advance, how many times has the close date been pushed. An opportunity pushed three times is no longer a timing question, it is a reality question. Measuring how close your forecast lands to actuals, period after period, makes that reading much faster; the method is in our article on forecast accuracy.

Timing variance also has a cash side, and it usually hits harder than the sales side. Even where revenue was recognized inside the period, a collection that slipped creates a squeeze the sheet never shows. Why those two calendars need tracking separately is in our piece on cash flow management.

Putting the cause into one of five boxes

Leave cause-hunting open-ended and every month produces a new story, none of which can be compared to the last. A fixed set of boxes speeds up the search and makes explanations comparable across periods.

Source of variance	Distinguishing question	Matching action
Demand	Where does opportunity count sit against plan?	Invest in channels and demand creation
Conversion	Did the same opportunity close at a lower rate?	Stage review and coaching
Price	Why did unit price move away from plan?	Discount authority and approval thresholds
Timing	Was the deal lost or did it shift?	Period-end close discipline
Plan	Was the target coherent to begin with?	Rebuild the underlying assumptions

How the boxes sit relative to each other carries information too. Demand intact but conversion down means the issue is in the work the team is doing; demand down with conversion flat means the issue is on the marketing and channel side. Measuring at the start of the period whether opportunity volume is even sufficient for the target makes that separation easier; we walked through the calculation in pipeline coverage ratio.

Favorable variances deserve explaining too

Almost every team looks only at the red. Yet a line landing clearly above target carries at least as much information as one falling short, and often more, because something repeatable may have been found. A team that does not examine it misses two things at once: the behavior worth copying, and the risk of mistaking a one-off event for durable capacity.

Favorable variances usually have one of three sources: a single large deal that happened to land inside the period boundary, a product mix richer than planned, or a genuine improvement in conversion. The three imply completely different things for next period. The first will not repeat, the second partly repeats, the third permanently raises the ceiling. A team that lifts the next target without separating them builds a quota it cannot hit; how to set quotas soundly is covered in setting sales quotas.

What if the target is the thing that is wrong?

Variance analysis carries a silent assumption: the plan is correct and the actual is what needs explaining. This is where the standard practice needs contradicting. A meaningful share of variances originate not in the actual but in the plan itself. When a growth rate handed down from above is distributed line by line across regions, you end up with targets that correspond to no region's reality, and the same variance repeats every month.

The diagnosis is easy: if a variance of similar size and the same direction repeats for three consecutive periods, the problem is not execution, it is planning. The right response is not more pressure on the team but opening up the plan's assumptions. What conversion rate, what average value, what capacity was assumed? A budget structure that keeps assumptions visible compresses that argument from three months into one meeting; we laid out the skeleton in budgeting for small businesses.

A practical rule falls out of this: do not change a target mid-period, but do update its assumptions mid-period. A target is a commitment and loses its meaning once it moves; an assumption is an estimate and produces bad management if it never moves. Teams that keep the two apart honor the commitment and still track reality. How to update a forecast inside the period is in our piece on sales forecasting.

The anatomy of a variance meeting

Set up badly, the variance meeting turns into a defense hearing and rarely recovers. Three things determine the setup: who speaks, what they bring, and how long it runs. The person presenting the number and the person explaining the variance should not be the same; the first comes from reporting, the second from whoever owns the work.

One rule alone changes the quality of the hour: arrive with example records, not with a figure. When the person explaining a price variance opens three opportunity records and shows at which stage, on whose approval and on what grounds the discount was given, the discussion becomes concrete. Explanations offered without opening a record turn into copies of each other within a few months: the market was quiet, the competitor was aggressive, the customer postponed.

The standing agenda

Three headings are enough. First, the variances above threshold and the dominant component in each. Second, a one-sentence cause hypothesis per variance plus the two records that support it. Third, actions with an owner and a date. Everything discussed outside those three headings lengthens the meeting and becomes a conversation nobody remembers a month later. How to tie the monthly variance rhythm to the annual calendar is in our article on annual sales planning rituals.

From variance to action: owner, decision, date

The output of variance analysis is not an explanation, it is a decision. An explanation records the cause; a decision states what will be done differently next period. Teams that confuse the two produce better and better explanations every month while continuing to see the same numbers.

When you write an action, fill three fields: who, what, when. We will improve discount discipline is not an action. Discounts above a set level route to sales manager approval, configured in the system on the fifteenth is an action. The first item on the next meeting's agenda should be the status of those actions; without that, the list quietly becomes a wish list.

To find out whether the action worked, you need a comparison baseline, and that baseline does not come from outside. The two sides of the date you rolled it out have to be read in the same slice with the same definition. How to cut a ruler out of your own data is covered in internal benchmarking.

Common mistakes and where to start

The most frequent mistake is looking at variance only at total revenue level; a difference that appears small in the total may be two large opposing variances hiding each other. The second is seeing the variance only at month end: a sheet read on the fifteenth still describes a period you can act on, while one read on the last day just records history. The third is attaching a variance directly to a person; unless the cause is attached to a component, the action gets stuck on the person and leaves no trace in the system.

To start, stay small. Pick one metric — revenue is enough — and build a single table that splits it into three components. Read only that table for two periods, cap explanations at two sentences, and write every action with a date and an owner. By the third period the table starts explaining itself, and that is when you open the breakdown to product and region. Multiplying components from day one is the fastest way to turn the table into a screen nobody reads.

Variance analysis is effortless where plan and actual are read from the same set of data, and turns into a monthly rebuild where they live in two separate files. In Rocketly, targets, opportunities, win and loss reasons, invoices and period reports all sit on the same structure, so decomposing a difference becomes a matter of a few filters — open a free account and build your own variance view.