Turn insights into a data-driven testing roadmap

1. Talk to stakeholders and look at traffic, funnel, acquisitions, landing page, and device type reports in Google Analytics to identify at least five trackable key metrics related to your business goals.

Compile past data for these metrics in a spreadsheet.

2. Use the analytics data to investigate visitor behavior and attributes such as which device they use, the site that referred them or their geographical location, and note commonalities.

For example, a substantial percentage of visitors to your site might be from a single country.

3. Create segments for your users based on the commonalities you identified.

4. Analyze your conversion funnel, filtering by the segments you created, to look for leaks and pain points that are affecting some of your audience.

For example, if you noticed that a lot of people are accessing your ecommerce landing page on their mobile devices. Using session recordings, you can find frustrations or hesitations that mobile users experience before they make a purchase or abandon your site. Use behavioral analysis tools such as heat maps and session recordings to identify leaks.

5. List the pages that need optimization in a spreadsheet and construct at least one hypothesis for each, combining where a hypothesis covers multiple pages.

Complete this statement to create and articulate a hypothesis: “I believe______, and if I am right then _______, because_________.” The latter part of the hypothesis is the qualification or the outcome that you expect and why you think so.  For example: I believe that adding customer testimonials to key pages will address the low purchase rate of our audience because they will act as social proof to establish the credibility of our product/service.

6. Sort your hypotheses in order of importance using the ICE (impact, confidence, ease) model and assign a date to the most important ones.

7. Set up your analytics or an integrated testing system with analytics and tag management tools like VWO to measure your KPIs against the current baseline and desired goal.

8. Run an A/B test to investigate a hypothesis and document the findings so that you can reference them in future testing and optimization exercises.

For example, to test the impact of social proof on your conversions, run an A/B test where your control is a landing page with no testimonials and the variation is the same, but with testimonials from existing or past customers.

9. Analyze your test results and either modify your hypothesis if the test failed, tweak your hypothesis if you didn't achieve significant results, or move onto the next hypothesis if the test was successful.

If your experiment was a success, move on to the next important metrics to test. If your test failed, document your learnings from the bad test, revisit your hypothesis, and start over. If it moved the needle but did not fulfill the goal, try another experiment by tweaking your hypothesis a bit.