There, I said it. A/B test hypotheses, as they're commonly written, are wasted keystrokes that don't do enough to ensure your test results will be valid or that your analysis will be accurate. Let's …
When a variation is losing big time, early into a test
You know better than to call a winner 4 days into a test, just because a calculator calls the results significant. But what about losers? The same statistical considerations apply. You need more …
Continue Reading about When a variation is losing big time, early into a test →
The peeking problem
We all know you shouldn't call an A/B test early, because it increases the chance that you'll mistakenly declare a winner, and that is arguably worse than not testing at all. The solution, then, is …
Dealing with pressure to call tests early (3/3)
In Part 1 we tried to see things from the perspective of the execs, clients, media buyers, and bystanders who might favor calling a test as soon as you see lift. They have budgets, and goals to hit - …
Continue Reading about Dealing with pressure to call tests early (3/3) →
Dealing with pressure to call tests early (2/3)
Yesterday we tried to get inside the heads and hearts of stakeholders who might favor calling a test early. Today, let's look at a few tools you can use to make the case for running tests to …
Continue Reading about Dealing with pressure to call tests early (2/3) →
Dealing with pressure to call tests early (1/3)
You already know that calling a test too early is a Bad Thing. It can lead to Type I and Type II errors, leaving your conversion rates worse than they could've been, or even worse than where they …
Continue Reading about Dealing with pressure to call tests early (1/3) →