What is it?
An A/B test is the closest that advertising comes to a true randomised controlled trial. We show adverts to some people and not to others, and then we track their behaviour, measuring the difference that the adverts make.
The trick to getting A/B testing right is working out beforehand how large the test needs to be, and how long it needs to run, to generate a reliable result. We also need to ensure that the selection of targeted and non-targeted people is random, or at the very least that we’re aware how it is not random, which in a world of algorithmically targeted advertising can be a significant challenge.
With a good test, you learn the true incremental impact of making a change to your advertising, gaining a robust data point to feed back into marketing mix modelling and budget setting.
When is it used?
- Creative, offer or landing-page tests
- Any time that advertising can be targeted 1:1 to specific people
Why Sequence?
- Deep statistical modelling experience to ensure tests are robust
- ‘No result’ is still useful. We design tests to be large enough so that you know if you have not measured an uplift, the activity did not meet your targets
- A full suite of measurement tools. Often testing is the right approach, but if it isn’t then we’ll tell you what’s better
- Test results feed back into marketing mix modelling and budget setting
Next service
Budget optimisation →