AOV × Orders Revenue Decomposition
A practical framework for splitting revenue change into AOV movement vs order-count movement — including the cross-term — so you know whether to fund basket-size levers or acquisition next quarter.
AOV × Orders Revenue Decomposition
A price × volume split of revenue change, isolating AOV movement, order-count movement, and the cross-term between them.
AOV × Orders Revenue Decomposition is the price-times-volume identity applied to an online store's top line: revenue equals average order value multiplied by order count, so any revenue change can be split into how much came from AOV moving, how much came from orders moving, and a small interaction term where both moved at once.
The framework matters because the two halves of the equation are funded by completely different work. AOV is a merchandising, bundling, and on-site UX problem. Orders is an acquisition, retention, and conversion-rate problem. Without the decomposition you end up arguing about budgets on vibes; with it you know whether next quarter's marginal euro should buy more traffic or lift basket size.
Revenue can only move in four ways: more orders at the same AOV, higher AOV on the same orders, both at once, or one of them moving enough to mask the other. The decomposition is just the arithmetic that makes those four cases legible on a single slide.
For a Shopify apparel brand growing 18% year-on-year, the headline is useless on its own. If orders are up 25% but AOV is down 6%, you have a discount-led growth problem dressed up as a win. If AOV is up 15% and orders are flat, your acquisition engine has stalled and bundles are carrying the quarter. Same 18%, three very different funding decisions.
The math: three terms, not two
Start from the identity: Revenue = AOV × Orders. Take the difference between two periods and you get three components, not two: an AOV effect (ΔAOV × Orders₀), an Orders effect (ΔOrders × AOV₀), and a cross-term (ΔAOV × ΔOrders) that captures the bit where both moved together.
Most operators silently drop the cross-term or assign it arbitrarily to whichever lever they want to celebrate. That's fine when the deltas are small — a 2% AOV lift on 3% order growth has a cross-term worth 0.06% of revenue — but it becomes dishonest in peak season, post-launch, or after a price increase, when both axes swing 10%+ and the interaction can be 1-2% of total revenue all by itself.
Reading the four diagnostic patterns
Pattern one — orders up, AOV flat — is the cleanest read: your acquisition or conversion work is paying off and nothing in the basket is breaking. Pattern two — AOV up, orders flat — usually means bundles, thresholds, or a price test are working, but you're not pouring fuel on a growing fire from the top of funnel.
Pattern three — orders up, AOV down — is the mix-shift trap. New customers, a paid channel skewed to entry-price SKUs, or a promo wave can lift orders while dragging average basket value down. Pattern four — AOV up, orders down — is what a price increase or a pruned discount calendar looks like before you've decided whether the trade was worth it.
The cross-term trap
If you allocate the cross-term entirely to AOV (or entirely to Orders), you'll over-credit whichever lever you're already biased to fund. The honest defaults are: split it 50/50, attribute it proportionally to each effect's size, or report it as its own line on the bridge. Pick one method and use it consistently across every quarterly review — flipping methods mid-year is how decomposition arguments turn into religious wars.
From decomposition to funding decision
The decomposition only earns its keep when it changes what you fund. If the AOV effect is the larger half of the bridge and the trend has held for two quarters, the marginal euro probably belongs in bundles, free-shipping thresholds, post-purchase upsells, or a serious look at PDP cross-sell. The acquisition team's case for more spend is weaker — they're already converting; the basket is the bottleneck.
If the Orders effect is the larger half, the case flips. Acquisition, retention email flows, and CRO work on the funnel get the next budget cycle, and AOV experiments move to the maintenance backlog. When you also break the decomposition down by new vs returning cohorts or by paid channel, you usually find one segment driving the whole bridge — and that's where the experiment roadmap goes.
How a €180k revenue lift attributes under three cross-term methods
AOV effect
Orders effect
Cross-term (kept separate)
AOV × Orders decomposition: common questions
It's the price-times-volume identity applied to ecommerce: revenue equals average order value times order count, so any change in revenue can be split into an AOV effect, an Orders effect, and a small cross-term where both moved together. The split tells you which lever is actually driving growth or decline.
The three defensible methods are: drop it when both deltas are tiny (under ~3% each), split it 50/50 between AOV and Orders, or allocate it proportionally to the size of each main effect. Pick one and stay consistent across periods — switching methods mid-year invalidates trend comparisons.
When the AOV decline is driven by discount depth, mix shift toward entry-price SKUs, or a paid channel skewed to bargain hunters — none of which compound. If the AOV drop is from genuine new-customer acquisition that you expect to repurchase at full price, it's a healthy investment phase, not a problem.
YoY removes seasonality and is the right view for board reporting and annual planning. QoQ catches recent inflections faster but gets noisy around peak season and product launches. Most teams run both and only act on a trend when YoY and the last two QoQ reads point the same direction.
A revenue waterfall typically breaks change into business-unit, channel, or product lines. AOV × Orders decomposition is one specific waterfall built on the price-volume identity. You can nest them — decompose total revenue by channel first, then run AOV × Orders inside each channel to find where the drag is.
Yes, and you usually should. Meta-driven orders often have a lower AOV than direct or email-driven orders, so a shift in channel mix can move blended AOV without anything changing on-site. Running the decomposition per channel surfaces which channel is dragging the average down.
New customers almost always have a lower AOV than returning customers, so an acquisition push will mechanically depress blended AOV. Decomposing by cohort separates 'we acquired more first-timers' from 'our returning customers spent less' — two completely different problems with different fixes.
The highest-leverage AOV moves are free-shipping thresholds set just above current AOV, post-purchase one-click upsells, bundle SKUs with a clear per-unit discount, and PDP cross-sell tied to genuinely complementary items. Price increases work too but compound with elasticity risk.
Compare gross AOV (before discount) against net AOV (after discount) for the same period. If gross AOV is flat or rising but net AOV is falling, you have a promo-depth problem, not a demand problem. The fix is the discount calendar, not the basket.
Rule of thumb: if both ΔAOV% and ΔOrders% are under 3%, the cross-term is under 0.1% of revenue and can be quietly absorbed. Above 5% on either side, surface it as its own line. Above 10% on both sides — peak season, post-launch — the cross-term can hit 1-2% of revenue and must be reported.
Test ideas before you ship them
Run unlimited A/B tests, attach hypotheses to outcomes, and build a searchable archive of what works — and what doesn't.