Price, Volume and Mix Variance Inside a KPI Tree

The Revenue Number Moved and Nobody Can Say Why

Price volume mix variance is the arithmetic that answers what a revenue chart cannot: of the change between two periods, how much came from charging more, how much from selling more, and how much from selling a different blend of things.

Most quarterly reviews stop one level short of this. The deck shows revenue down 1 percent. One person says demand was soft. Another says a large account renegotiated. Both are stories. Neither is measured.

The gap is rarely analytical talent. The decomposition simply never gets built. FP&A Trends, in its 2025 benchmarks read on August 11, 2026, reports that 77 percent of organizations running dynamic or fully driver-based models rate their internal forecasts as good or great, against 27 percent of organizations using basic models or none.¹ The 2026 AFP FP&A Benchmarking Survey on integrated planning, drawing on 332 finance professionals across 54 countries, examines the same adoption gap.²

A price volume mix bridge is the smallest driver-based model worth owning. It needs two periods of transaction lines and returns three numbers that add up to the variance you are trying to explain.

What Is Price Volume Mix Variance?

Price volume mix variance splits the change in revenue between two periods into three additive terms: a price effect, a volume effect, and a mix effect. The three sum exactly to the total revenue variance. Because they add rather than multiply, they sit naturally as sibling nodes beneath a revenue variance node in a KPI tree such as kpitree.io.

Price effect isolates what changed because you charged a different amount for the same thing. Volume effect isolates what changed because total units moved. Mix effect isolates what changed because the composition of those units shifted toward cheaper or more expensive items.

The vocabulary is older than the software. Management accounting has carried the same idea as sales price variance, sales mix variance and sales quantity variance for decades. Martin's Management Accounting Textbook, chapter 13, states the governing identity plainly: the sales mix variance plus the sales quantity variance must equal the sales volume variance.³ AccountingTools defines sales mix variance as the change in profit or contribution attributable to variation in the proportion of products sold away from the standard mix.⁴

Consulting practice uses the commercial names. FTI Consulting's white paper on a quantifiable approach to price volume mix analysis frames the same three terms for revenue and gross margin bridges.⁶

Whichever vocabulary your organization uses, the structural property is the one that matters: the terms are additive, and the sum is a hard constraint you can test.

The Three Terms, Written Out

Write P for unit price and Q for units. Subscript 0 is the base period, subscript 1 the current period. The index i runs over whatever dimension you sold across: product, SKU, customer, region, or channel.

Price effect = sum over i of (P1i - P0i) x Q1i

Volume effect = (Q1 total - Q0 total) x P0 average

Mix effect = sum over i of (P0i - P0 average) x (Q1i - Q1 total x S0i)

Three definitions carry the weight. P0 average is base period revenue divided by base period units, not the average of the individual prices. S0i is item i's share of base period units. Q1 total x S0i is therefore the current period unit count item i would have sold if the mix had not moved.

Read the mix term in plain language. For each item, take how far its price sits above or below the base average, and multiply by how far its unit count sits above or below its mix-neutral count. Selling more of a below-average item produces a negative contribution. Selling more of an above-average item produces a positive one.

A Worked Example That Reconciles to the Dollar

Two products, two quarters, no rounding.

Quarter one: product A sold 1,000 units at 10 dollars, product B sold 1,000 units at 30 dollars. Total 2,000 units and 40,000 dollars. Base average price is 40,000 divided by 2,000, which is 20 dollars.

Quarter two: product A sold 1,400 units at 10 dollars, product B sold 800 units at 32 dollars. Total 2,200 units and 39,600 dollars.

Units rose by 200. Product B's price rose by 2 dollars. Product A's price held. Revenue still fell by 400 dollars.

Price effect: A contributes (10 - 10) x 1,400 = 0. B contributes (32 - 30) x 800 = 1,600. Total plus 1,600 dollars.

Volume effect: (2,200 - 2,000) x 20 = plus 4,000 dollars.

Mix effect: base shares are 50 percent each, so the mix-neutral split of 2,200 units is 1,100 and 1,100. A contributes (10 - 20) x (1,400 - 1,100) = minus 3,000. B contributes (30 - 20) x (800 - 1,100) = minus 3,000. Total minus 6,000 dollars.

Add them: 1,600 plus 4,000 minus 6,000 equals minus 400. That is the reported variance, to the dollar.

The reconciliation is the test, not a formality. Price effect plus volume effect plus mix effect must equal current revenue minus base revenue. If it does not, the bridge is wrong, and no narrative built on it is safe.

Why Can Revenue Fall When Units and Prices Both Rise?

Because volume and mix are different things. Total units can rise while the blend shifts toward cheaper items, and the mix term captures that shift alone. In the worked example, mix costs 6,000 dollars while volume and price together add 5,600. The business sold more units of a less valuable product.

This is the failure mode that makes two-term bridges dangerous. A price and volume split with no mix term buries the composition shift inside the volume number, and the volume number then reads as good news.

The commercial consequences are specific. A sales team compensated on units will grow units and shrink revenue. A discount program aimed at a low-price entry product will look like demand strength. A channel expansion into a lower-priced format will read as growth until the average selling price is examined separately.

Mix is also where most double counting happens. If the same shift is allowed to appear in both the volume term and the mix term, the bridge over-explains and the residual has to be plugged. A plugged residual is the clearest sign that the decomposition is not mutually exclusive.

The discipline that prevents this is the same discipline behind the three decomposition patterns in the KPI tree template: every child must be defined so that no dollar can be counted under two parents.

Where the Three Terms Belong in the Tree

Put the variance at the root of its own subtree, not inside the revenue tree.

A revenue tree holds levels: revenue equals the sum of segment revenue, segment revenue equals the sum of product revenue, and so on. A variance tree holds changes. Its root is current revenue minus base revenue, and its three children are the price, volume and mix effects.

Below each effect, decompose by the same dimension you indexed over. The price effect splits into per-product price effects that sum to it. The volume effect splits into whatever drives units: sessions and conversion for e-commerce, accounts and usage for subscription, distribution points and rate of sale for consumer goods. The mix effect splits into per-item mix contributions.

Keep the two trees adjacent rather than nested. Analysts use the level tree to answer what the business looks like, and the variance tree to answer what changed. Merging them produces a structure where some nodes are stocks and some are deltas, and the arithmetic stops being checkable.

The worked KPI tree examples on kpitree.io cover the level side. This article covers the variance side.

Two-Way, Three-Way and Four-Way Conventions

There is no single correct decomposition. There is a family, and the members differ in how they treat the interaction between price and quantity.

The interaction term is real. When both price and volume move, part of the revenue change belongs to neither alone. Two-way splits assign it by convention. Three-way splits usually absorb it into the price term by valuing price at current volumes, which is why the formula above multiplies the price delta by Q1 rather than Q0. Four-way splits report it separately.

Pick one convention, write it into the metric definition, and do not change it mid-year. Comparability of a bridge across quarters matters more than the theoretical elegance of any single variant. CIMA's P1 syllabus treats sales mix and sales quantity as a further split of the sales volume variance, which is a fourth way of arranging the same arithmetic.⁵

The table below states what each convention produces and when it is the right choice.

How Do You Test a Price Volume Mix Bridge for Correctness?

Run four checks. The three terms must sum to the reported revenue variance. Each term must be computable from summable columns only. Base average price must come from total revenue divided by total units. And every item must appear in both periods, with absent items entered as zero units rather than dropped.

Test 1. Reconciliation. Sum the three effects and compare to current revenue minus base revenue. A gap of any size means the bridge is broken, not approximately right.

Test 2. Summability. Every input must be a column you can add: revenue and units. Price is derived by division, never stored per row. A dataset that only carries an average price column cannot produce a correct bridge, because averages do not add.

Test 3. Base average integrity. Compute the base average price as total base revenue divided by total base units. Using the mean of per-product prices weights a 10-unit product the same as a 10,000-unit product and silently corrupts the mix term.

Test 4. Entry and exit. Products launched or discontinued between periods break naive joins. Enter them with zero units and zero revenue in the missing period. A product that appears only in the current period should show its full revenue across the volume and mix terms, not vanish from the total.

Failing any of the four turns the bridge into a plausible-looking narrative with no arithmetic underneath it.

Where a Price Volume Mix Bridge Is the Wrong Tool

A price volume mix bridge explains a revenue variance. It does not establish causation, and it is a poor instrument in four situations.

Single-product businesses. With one item, the mix term is zero by construction. A price and volume split is sufficient and easier to read.

Contract revenue with no unit. Professional services, licensing and usage-tiered subscriptions often lack a clean unit. Decomposing by hours, seats or consumed credits works, but the mix term then describes contract shape rather than product blend, and needs relabeling.

Very short periods. Week-over-week bridges on volatile categories produce large mix swings that reverse the following week. The arithmetic is right and the signal is noise.

Attribution questions. The bridge says mix cost 6,000 dollars. It does not say whether that came from a promotion, a stockout, a competitor launch, or a change in channel weighting. That is a separate question, answered by decomposing the mix term further or by repeatable root cause analysis on the affected branch.

Naming the limits is part of shipping the method. A bridge presented as causal proof will not survive its first serious challenge.

Building the Bridge in kpitree.io

kpitree.io is a self-service KPI tree builder. This section is the only part of the article about the product.

The mechanics line up with the four tests above. You upload a CSV of transaction lines carrying revenue and units per item per period. Inside the tree, derived KPIs are computed by dividing two summable columns, so average price is defined as revenue divided by units at every node rather than stored per row. That is exactly what Test 2 and Test 3 require.

Node identities in the tree are addition and subtraction. A variance node whose children are the price, volume and mix effects is therefore a native structure, and the reconciliation in Test 1 becomes a property of the tree rather than a manual check in a spreadsheet.

Two honest limits. The evidenced ingest path today is CSV upload, so a bridge is refreshed by uploading a new file, not by a live connection. And capabilities the site marks as coming soon, including natural language controls, are not available.

The analysis in promotion effectiveness in CPG applies the same arithmetic to a trade promotion question.

Frequently Asked Questions

Is price volume mix the same as sales variance analysis?

They are the same arithmetic under different names. Management accounting says sales price variance, sales mix variance and sales quantity variance. Commercial finance says price, volume and mix. Martin's Management Accounting Textbook and CIMA's P1 syllabus both treat mix and quantity as a split of the sales volume variance.³ ⁵

Can I run a price volume mix bridge on gross margin instead of revenue?

Yes. Run one bridge on revenue and a second on cost of goods sold, using the same item index and the same convention, then subtract. FTI Consulting describes this combination as the standard route to a gross margin bridge.⁶

What if a product has no price?

Then it has no place in the price term. Give it a unit, or exclude it from the bridge and reconcile it as a separate line. Do not assign it an assumed price.

How many items should I index over?

Ten to thirty is typical. Below ten, mix rarely explains much. Above a few hundred, aggregate to a product family first, then drill into the family that moved.

Does the order of the terms change the answer?

It changes how the price and volume interaction is allocated. It does not change the total. That is why the convention has to be fixed once and documented.

Closing: Rebuild Last Quarter's Revenue Bridge Before the Next Review

The next time revenue moves and the review turns into a debate, the fastest way to end it is a bridge that reconciles.

Take one CSV. Two periods of transaction lines, one row per item per period, with a revenue column and a units column. Nothing else is required. Compute the base average price as total revenue over total units, run the three formulas, and check that they sum to the reported variance.

If they do, you have a defensible answer in an hour rather than a narrative in a meeting. If they do not, you have found a data problem worth more than the answer would have been.

Start with a single product family and a single quarter. Upload that one CSV and decompose the metric your team already argues about.

kpitree.io is a self-service KPI tree builder for finance, business and product analysts.

Sources

• 2025 FP&A Benchmarks: Where We Are, Where Leaders Are Going
• 2026 AFP FP&A Benchmarking Survey Report: Integrated Planning
• Martin's Management Accounting Textbook, Chapter 13
• Sales mix variance definition
• CIMA P1 Notes: Sales Mix and Quantity Variances
• A Quantifiable Approach To Price Volume Mix Analysis