← DictionaryA/B Test
also called split test
Comparing two versions against one variable to see which performs better.
Two versions, one difference, measured.
What It Is
A/B Test is a Measurement item, filed under Testing. The canonical meaning is comparing two versions against one variable to see which performs better.
It is a repeatable structure, which means it only pays off when it is actually followed through instead of quoted. That definition is deliberately narrow — the narrowness is what makes it usable.
How It's Used
A/B Test earns its keep in reporting, analytics reviews and the arguments about what the numbers mean. It rarely travels alone — expect it next to Statistical Significance, Conversion Rate Optimization and Baseline. The practical test is whether anyone can follow it without the person who wrote it in the room.
Why It Matters
The cost of being vague about A/B Test shows up later, not today. When it is unclear, reporting becomes a matter of opinion. The businesses that treat it seriously spend less time redoing work.
It is a set of steps someone worked out so you do not have to guess the order. Two versions, one difference, measured.
Termshift™
How the meaning shifts depending on who is using it, and where.
In agency and in-house rooms, A/B Test means exactly comparing two versions against one variable to see which performs better.
In plain conversation it comes out as two versions, one difference, measured.
The Bottom Line
A/B Test pays off when it is specific and costs you when it is not. Two versions, one difference, measured.