##### Preferences
• Metric: Average Order Value
• Statistics: Z-Test
• Tails: 2
• Confidence: 95%
• Power: 80%
• SRM Confidence: 99%
##### Data Input
To compute AOV, the CSV file should have 1 variant per column, 1 order value per cell, and no headers. If you need an example, you can download demo-aov.csv. The 1st column always corresponds to the control variant, and the only column needed for test planning. In that latter case, you need to include one-week worth of values!

Pre-Test
Values per week
###### Variants
No variants present, pre-test mode

# Average Order Value Calculator

AOV is a fine eCommerce metric to monitor but it’s a little more delicate to optimize, for the simple reason that its randomization unit is actually the orders (conversions) themselves, and not the visitors.

$AOV = \dfrac{Revenue}{Conversions}$

Meaning that, if we want to compare apples to apples, we need all variants to maintain the same conversion rate! The kind of things that can be done to increase order value is tactics like upsells and cross-sells, even increasing product prices is risky as it can cause the conversion rate to move.

As a rule of thumb, if you are just looking to optimize revenue, go with ARPV. Now, if you know what you are doing, please go ahead and have fun with AOV.

This calculator allows you to do both pre-test and post-test for average order value.

For the pre-test, you can calculate the MDEs (Minimum Detectable Effects) by using a CSV file with one week’s worth of order values, without null orders in one column.

As for the post hoc (post-test) analysis, you need to have all non-zero order values, one column per variant, with the first one always being the control. The values are order values without 0 values, and for the full duration of the test, or up to the present moment if the test is still running.

AOV is very important to observe as a metric because even though we optimize for conversion rate, AOV itself could go down and our overall bottom line could take a hit as a result. So it’s a good guardrail metric to observe.

However, it’s tricky to test, as its randomization unit is the conversions themselves. So during an A/B test involving AOV, it’s important to keep the conversion rate equal for all variants, which means the page should be the exact same. Only things like up-sells or cross-sells should be attempted (even those can influence the CR, so it’s really delicate).

As a rule of thumb, if you need to do revenue optimization, go with ARPV unless you really know what you are doing!

While the Average Order Value is not the most complete eCommerce optimization metric, it is still regularly monitored by site owners and A/B testers.

Average Order Value is an indication of the profitability of an eCommerce shop, across segments. It’s a rough and ready, broad metric that quantifies how much people buy across categories.

However, when using it as the primary eCommerce metric or KPI, the following things need to be kept in mind:

• Average order values must also be monitored across categories.
• Orders from distributors or those who purchase in bulk can throw this value off. Be vigilant.
• The average order value will not indicate profit. It does not subtract the COGS (Cost of Goods Sold) which is a significant expense.