Typical CRO process

You just started working on a website to improve conversions. Where to start?

The typical rookie mistake is using the spaghetti method – start throwing stuff against the wall, see if it sticks. People start with random A/B testing, and then wonder why no gains were achieved. This kills the mood and the momentum.

Instead, follow a step-by-step procedure. Conversion optimization is a process and should always be described as such. Here’s how it goes:

1. Determining business objectives: deciding which metric(s) we want to increase. 
2. Business Questions: Come up with as many questions as possible from the user’s perspective
3. Data gathering: Set up relevant data collection and tracking. Wait for the data to come in.
4. Insight phase: Analyze all the different sources of data, identify patterns, draw conclusions.
5. Identify Problems: Refer to the data to discover problem areas which need attention.
6. Hypothesis: Develop proper hypotheses for tests. Rank the hypotheses based on their potential and ease of implementation. This determines which ones we’d test first.
7. Design: Based on the hypotheses you develop wireframes for treatments. Wireframes are mainly used as communication and planning tool. Once completed, wireframes will be passed on to the designer who will turn them into design screens (alternative layouts / content for pages on your site).
8. Technical integration: This is a necessary evil – new design screens need to be integrated onto the website. If the changes are minimal, this can also be done via testing tool itself.
9. Testing: We do proper controlled a/b testing following the scientific method, run every test until statistical significance.
10. Post-Test Analysis: Assess data to discover how different variations performed against goals.
11. Learning and improving customer theory: We analyze the test results (when done right, each test is a source for learning), gain new insights, develop new hypotheses. With each test our customer theory should improve.
12. Documentation: Archive your test results and learnings.

Back to step 2.

This is what I call a conversion cycle.

How long does one cycle take?

There’s no one answer for it. It can take 2 weeks, a month, or 6 months. This depends on

1. the complexity of the site (simple landing page vs huge e-commerce site),
2. the data we have available (how much still needs to be gathered, how long it takes to crunch through it),
3. and website traffic (this determines how long the test will have to run to achieve statistical significance).

The faster you can go through the cycle – without compromising quality of the work – the better. The more experiments you can run within 12 months, the more learning and potentially more revenue gains.

When does conversion optimization end?

The correct answer is ‘never’. You can always do better than you did last month.

The world is always changing, your business is evolving, as is your competitive landscape. New technologies, you name it. Change is the name of the game, and you need to keep up. Conversion optimization is part of it.