Conversion optimization is a continuous process and a long-term commitment. This is why you should not only look at the uplift of single A/B-tests but rather focus on the cumulative uplift of all tests.
To be successful in the long run, you need to understand the factors that are impacting your optimization work and how you can influence them.
There are three conversion optimization factors which will impact your business’ long-term growth:
- The number of tests you run
- Win rate (% of winning tests)
- Average uplift per test
How to improve all of these factors?
1. How to increase the number of tests
Establish a testing culture in the entire company.
The entire organization must embrace a mindset where testing, failing and learning is accepted as part of an iterative, agile procedure. It helps to have everyone on board at an early stage (the marketing department, sales, developers, customer support, and so on), e.g., during the idea-generation phase, for resource planning, and test idea input. Additionally, test results should be shared (and celebrated) with all stakeholders on a regular basis.
Most successful clients have a small, powerful, cross-functional team
One representative from each team is best. It helps with smooth planning and execution.
A cross-functional team can better react to resource bottlenecks; for instance, if resources are currently limited in the development department this can be reflected in the testing roadmap by giving priority to tests that need hardly any programming (such as simple tests where pictures or text elements are exchanged). If the designer is the bottleneck, tests with no need for graphic work can be pushed forward (such as the optimization of forms).
Run multiple separate tests at the same time
Just make sure that the tests are for different parts of the funnel, or for different segments. Don’t run two tests for a checkout page. Run one for checkout, and another for a newsletter sign up page. Or one for mobile, another for desktop. One for prospects, another for customers.
2. How to improve the win rate
Sound user research is the only solid basis for long-lasting higher conversion rates.
Best practices are nice and may be helpful but experience has shown repeatedly that optimization based on insights from user research delivers much better results. Without exceptions.
Test with a clear hypothesis in mind
This way you know what you want to test, why you want to test, and what results you expect. Seems trivial but you’d be surprised how often tests are run only for the sake of testing, with no clear purpose other than “I’ve seen it on X site and we should test this too.”
Here’s a good template:
Test ideas should be sorted by estimated uplift potential and implementation effort.
Evaluate the uplift potential and implementation effort with a number from 1 to 3. High uplift potential is rated a 1; likewise, low complexity is rated a 1. By multiplying the 2 numbers, you can easily rank your test ideas:
3. How to get higher uplifts
Start with research. Always.
Let’s reiterate this again: the better insights you gain from user research, the better you can formulate a strong hypothesis and achieve superior results with your tests.
Make your tests big, bold and prominent.
The more obvious and eye-catching the changes you are testing, the bigger their potential impact on conversions.
Focus on the “main roads”.
A quick analysis of your analytics data will reveal the major traffic pathways of your visitors. This will give you insights on the pages where tests will most likely have the biggest impact on your business.
PS. Pssst! Do you want me to tell you a secret? Do you want to know what’s the biggest obstacle to more testing?
It’s implementation (or development, or programming, or however else it is called in your company 🙂)
Without exception. Every company I ever worked with, from Amazon to one-person businesses, it is always implementation.
So if you want to make sure you get the most out of your conversion optimization efforts, make sure you specifically dedicate development resources for this.