Uplift

Term used to designate themeasurable increase in performance (conversions, revenue, engagement, etc.) generated by a specific action, compared with a reference situation. In A/B testing, the uplift represents the difference in results between the variant tested and the control group, and makes it possible to evaluate the real impact of a modification on key KPIs.

In CRO (Conversion Rate Optimization), uplift is a central indicator for quantifying the incremental value of an experiment.

Examples:

• If variant B of a button (e.g.: different color or wording) generates a conversion rate of 6% versus 5% for version A, the uplift is +20%.
• If a new version of the shopping cart page increases average revenue per visitor (RPV ) by €12, compared with €10 previously, the uplift is €2, i.e. +20% added value.
• In a test on a personalized recommendation feature, an increase in click-through rate from 3% to 4.5% represents an uplift of +50%.

The uplift can be expressed in :

• absolute value (e.g. +€2 RPV),
• relative percentage (e.g. +20% conversion),
• or estimated monetary value at traffic level (e.g.: +15,000 € / month).

Please note: an uplift is only significant if it is accompanied by sufficient statistical confidence, to avoid drawing conclusions based on variations due to chance.

‍

‍