What is average order value?

Why does the mean mislead?

Because order value distributions are skewed, and the mean follows the tail.

A store with 99 orders at $50 and one at $5,000 reports an average of $99.50. No customer spent anything near that. The median is $50, which describes the typical order honestly, and the mean describes a customer who does not exist.

Two practices that fix it:

Report the median alongside the mean. A wide gap between them signals a skewed distribution and tells you the mean is following outliers. Watch how the pair moves: AOV rising while the median holds flat means the tail moved rather than the typical customer, and the mean can sit still while the distribution changes underneath it.

Segment before averaging. By product category, traffic source, new versus returning, and device. A blended AOV across a $30 accessory line and a $2,000 equipment line describes neither.

For B2B SaaS the equivalent metric is average contract value, and the skew is worse rather than better. One enterprise deal landing in a month of self-serve signups moves the mean far enough that the reported figure describes no customer in either group, which makes the median and the segment split more load-bearing there than in retail.

Which levers raise AOV?

Ordered by return and by how well they survive contact with customers.

Order bumps and relevant cross-sells at the point of purchase. Small, genuinely related additions. Effective and low risk.

Bundling. Packaging related items at a modest discount raises the transaction and the perceived value together.

Free shipping thresholds. Set above current AOV, and effective because the incentive is concrete. Set too far above and it is ignored.

Tiered volume pricing. Works where buying more is genuinely useful, and backfires where it is not, since a discount on a quantity nobody needs reads as a markup on the quantity they do.

Upsells to a higher tier, presented as a comparison rather than a push.

Price increases, which raise AOV directly and belong in the conversation because they are frequently the largest available lever and the one nobody proposes.

What does raising AOV cost?

Every AOV lever risks conversion rate, which is why AOV should never be optimized alone.

The failure patterns:

• Aggressive upsells at checkout raise AOV among completers and raise abandonment, so total revenue can fall. Guard with cart abandonment.
• Bundles that remove the cheap entry point deter first-time buyers.
• High free-shipping thresholds raise AOV among those who chase them and lose those who do not.
• Cross-sell clutter at checkout adds decisions where the goal is completion.

Set revenue per visitor as the primary metric before running any of these, and report conversion rate and AOV underneath it as the mechanism. A test that raises AOV 20% and cuts conversion 25% is a loss that an AOV-only report would celebrate.

Related terms

Revenue per visitor · Cart abandonment · Checkout optimization · Guardrail metrics · Anchoring bias

Service: Conversion Rate Optimization.

FAQ

What is a good average order value?

Category decides it, so an outside figure has nothing to say about your catalog. Track your own trend segmented by product line and traffic source, and read the median alongside the mean so that a handful of large orders does not get reported as growth.

Should you use mean or median order value?

Report both. The mean drives revenue arithmetic and is sensitive to outliers. The median describes the typical order. Using the mean alone on a skewed distribution produces confident and wrong conclusions.

Does raising AOV always increase revenue?

No. AOV and conversion rate move against each other on most levers, so the question is whether the AOV gain outruns the conversion loss. The arithmetic is worth doing before the test rather than after: a 20% AOV gain pays for a conversion drop of roughly 17%, and anything past that is a net loss no matter how good the AOV chart looks.