Skip to content

Incremental revenue

E-commerce metrics

Incremental revenue is the revenue that would not have happened without a given action, such as an AI Shopping Assistant. It is measured with an A/B test against a control group.

How it works

The A/B test is the reference method.

  • Storefront visitors are split at random into two groups.
  • Group A sees the tool being tested, for example the AI Shopping Assistant. Group B, the control group, does not.
  • After a long enough period, you compare revenue per visitor in the two groups.
  • The gap, scaled to total traffic, gives the incremental revenue.

Because both groups are drawn at random, the gap can only be explained by the tool being tested.

Why it matters in e-commerce

It is the metric that answers the question asked of any investment: what did it bring in on top? It lets you calculate a real return on investment, by comparing incremental revenue with the cost of the tool. It also avoids crediting a tool with sales that would have happened anyway.

Real-life example: Samsonite

Samsonite measured the effect of the iAdvize AI Shopping Assistant with an A/B test over three months. The result was a 2.2x return on investment, driven by both higher conversion and higher average order value. With that proof, the brand scaled the assistant to 12 sites across its five brands.

 

  • 2.2x

    incremental ROI at Samsonite, A/B test over 3 months

  • 12

    Samsonite sites now running the AI Shopping Assistant

  • +15%

    incremental revenue measured by iAdvize customers

Samsonite: A/B test results. +15%: AI Shopping Assistant page.

Common confusion

Incremental revenue is often confused with AI-influenced revenue. AI-influenced revenue counts all sales from shoppers who interacted with the AI, including those that would have happened without it. Incremental revenue only keeps the difference the AI made. See: AI-influenced revenue.

The iAdvize approach

Measuring what the AI Shopping Assistant really brings in

iAdvize customers measure the effect of the AI Shopping Assistant with an A/B test, against a group of shoppers who do not see it. Shopper Insights then tracks the AI's contribution to the storefront's revenue.

 

Frequently asked questions

How long should an A/B test run?

Long enough to cover weekly traffic swings and gather enough orders in each group. Several weeks are often needed; Samsonite tested over three months.

How is it different from ROI?

ROI compares incremental revenue with the cost of the tool. An ROI of 2.2 means the tool brought in 2.2 times what it cost.

Can I measure incremental revenue without an A/B test?

You can estimate it by comparing two periods, but seasonality, promotions and traffic skew the result. The A/B test remains the only reliable measure.