Lift
Definition
Lift describes the relative or absolute improvement of a metric compared to a reference or control group.
Depending on context, lift can be defined differently statistically, marketing-wise, or model-wise. Without a clear baseline, the value is not interpretable.
Simply explained
How to measure how much better one variant performs compared to the comparison.
Why it matters in practice
Relevant for campaigns, models, and A/B tests.
Example from business
With equally sized comparable groups, the control group achieves 100 and the test group 110 conversions. This corresponds to 10% relative lift; uncertainty and experimental design must be examined separately.
Limits and pitfalls
- Not stating baseline; mixing relative and absolute increases; measuring attribution instead of causal lift.
Origin & history
Lift is a general comparison concept in statistics, marketing, and data mining.
Use in mid-sized companies
- Possible application in SMEs: Always report improvements with clear reference and uncertainty.
Frequently asked questions
Is 10% lift absolute or relative?
That must be defined. Does lift prove causality? Only with appropriate experimental design.
Sources
- Conversion Lift · Google Ads, Conversion Lift
Sources checked on 10.10.2026.
Editorially reviewed on: