WEBVTT

00:00:00.000 --> 00:00:06.000
How do you know whether a website change improves the result?

00:00:06.000 --> 00:00:13.000
An A/B test compares variations using a planned audience split and outcome measure.

00:00:13.000 --> 00:00:18.000
Plan the comparison before looking at the results.
6 A/B-testing basics.

00:00:18.000 --> 00:00:30.000
Start with a hypothesis
Explain what change might help and the outcome you expect.
Observed problem · Proposed change · Expected effect

00:00:30.000 --> 00:00:42.000
Define the variations
Keep the comparison focused enough to interpret.
Original version · Test variation · Defined difference

00:00:42.000 --> 00:00:54.000
Choose the outcome measure
Agree how success will be measured before starting.
Primary outcome · Useful guardrail · Reliable measurement

00:00:54.000 --> 00:01:06.000
Split the audience fairly
A planned allocation helps make the variations comparable.
Defined allocation · Comparable groups · Consistent experience

00:01:06.000 --> 00:01:18.000
Plan enough evidence
Sample size and run duration matter; a few visits can be misleading.
Required sample · Planned duration · Low-traffic limits

00:01:18.000 --> 00:01:30.000
Interpret the uncertainty
A small apparent difference does not automatically establish a winner.
Observed difference · Statistical uncertainty · Practical importance

00:01:30.000 --> 00:01:39.000
Your checklist
Start with a hypothesis · Define the variations · Choose the outcome measure · Split the audience fairly · Plan enough evidence · Interpret the uncertainty

00:01:39.000 --> 00:01:47.000
Treat an inconclusive result as inconclusive and choose the next useful investigation.
jerseyweb.je