Testing hypothesis creation got you down? Learn how to create winning ones

Do you know why most experimentation programs fail?

It’s most likely not what you think..

It’s not the complexity of the tests or the number of tests per month..

Most experimentation programs fail because they start from the wrong insights and the wrong testing hypotheses. If you’re building your tests solely on ideas that you got from a blog or a competitor or a list of “30 ideas for your online store” you might as well redesign your website blindfolded as you’re just as likely to get positive, sustainable results.

Here is a step-by-step guide into our own framework that you can use to generate solid hypotheses and have a much better chance

To be able to generate valuable insights, you need to start with asking the right questions:

❓What is the objective of our experimentation program?

❓What metrics do we focus on for the next 1-3 months?

❓What is the #1 obstacles in this conversion path for our users?

❓What is the #1 motivation for our users to convert?

❓Can I back this up with both qualitative and quantitative data?

The main objective of your hypothesis is to offer learning and the secondary objective is to increase/decrease a specific metric.

Validate your testing hypothesis by answering these questions:

❓How are you backing up your hypothesis? Do you have qualitative and quantitative data behind it?

❓What is the core principle behind the hypothesis? → Whether we’re talking UX, persuasion, functionality or anything in between, I think any hypothesis needs to revolve around a principle (reducing friction, creating reciprocity, improving visual hierarchy, etc)

❓What is the main metric you are using to analyse your test? What are the secondary metrics?

❓What defines a winner? Are you looking for an improvement in the main metric or for the variation to not decrease the main metric (testing adding a new feature and you want to make sure it doesn’t hurt the website performance).

You can’t do good CRO with just a list of tactics and stuff you got from one course, you need to have a process and to understand what you’re doing and, most importantly, why you are doing it.

If you’re looking for a more complex guide to creating testing hypotheses, please check out this article on the Convert.com blog – it will guide you through both the research and ideating process.

Don’t be afraid to think of testing copy – you’d be surprised at how much impact copy has on your conversion rate. We have a nice little guide here that can help you in crafting converting copy.

I’d be curious to find out how you validate your testing hypotheses before sending them to development so let me know in the comments.