Decide what to test next

Contributors

@paul-boag


Business Benefits

Make better choices to improve funnel conversions.


Talk to your stakeholders about your testing program’s goals, the metrics you want to track, the types of tests you’ll run, and the results you expect to achieve.

Make a list of your sales funnel pages, such as Homepage, Category page, Product page, and Cart.

Use Google Analytics to check metrics and dimensions that can help you identify the funnel pages with the highest potential for optimization.

Metrics:

  • Bounce rate
  • Conversion rate
  • Page value
  • Page load speed
  • Avg. Time on Page
  • Dimensions:
  • Source/Medium
  • Device category
  • Screen resolution
  • Campaign

Pages that users deem as important to their purchasing decisions (like Category pages and Product Detail pages) tend to have more potential to generate gains in revenue than those in the funnel’s later stages. Look for pages in your analytics that are performing badly based on the above metrics and target these for optimization. Target pages that perform badly based on the dimensions above.

Use heatmaps and session recordings from a tool like Hotjar to identify specific tests that you can run.

  • When watching back sessions or looking at heat maps, do users miss critical content or calls to action?
  • Are they having trouble filling in forms or making other mistakes?

These offer one or more potential improvements you can test.

Choose a page to optimize. Start a new spreadsheet and write down each test idea you have for that page in an Ideas column.

  • Choose the page that is experiencing the biggest problem and which has the greatest opportunity for improvement.
  • Include ideas and possible improvements that could be made. For example, moving a call to action or amending the copy of a headline.

Add a column for each test prioritization criterion that you want to evaluate.

Include criteria like:

  • Impacts an element above the fold
  • Promotes a change that can be noticed within 5 seconds
  • Adds or removes an element
  • Designed to increase user motivation
  • Will run on a high traffic page
  • Addresses an issue found via user testing
  • Addresses an issue found via qualitative feedback
  • Addresses an issue found via quantitative data or digital analytics
  • Addresses issues supported by click/scroll heatmaps or eye-tracking
  • Can be implemented in less than 2 days

For each test idea, answer each criterion with Yes or No.

Add a column titled Score, add up the number of Yes answers in each row, and enter the result in this column.

The ideas with the highest scores should be tested first.

Start testing by implementing the ideas with the highest scores as A/B tests - track insights in GA or any other analytics platform that you use.

For optimal insights, run tests for at least 2 weeks to decrease costs and get the right amount of data to achieve statistical significance. If you are testing on a high traffic page, you can implement more than one idea as a variant in each test. Use an A/B test calculator and select Pre-Test Analysis to verify how many variants you can afford to run with your current traffic.

Create reports for your stakeholders based on your test results, including KPIs such as conversions, click-through rate, and revenue per visitor.

In the report, add the process of deciding what to test using website analytics, the set of criteria from your spreadsheet and funnel performance to increase transparency.

After you run the first sequence of tests, consider repeating the process with a new selection of ideas.