Experiments (Split testing)

What is split testing and why is it important?

Important note: This is not a feature of the profile website - Learn about upgrading your website.
Currently it is also not possible to run split tests on product / shop pages. For example the category page, product page, shopping cart page and the checkout page. It is much more complicated to run such tests. If you are serious about tuning these pages, please speak to Zeald support.

Once you have decided what page you want to test and Setup the goals that you will measure the success by:
  1. Duplicate the page that you want to test using the duplicate link in the content tree.
    duplicate-page
  2. It is a good idea to store these test pages inside a dedicated top level category called tests. Call the page something like "test 1"
    test-pages
  3. make the changes to the new page. Use our optimisation checklist for ideas to optimise your page. Avoid using multiple components otherwise you will make it more difficult to port across the changes if the new page proves to be a winner
  4. Under the marketing tab, select the experiments tab to view a list of all your past experiments
    experiments1
  5. Select the Add a new experiment button to create a new experiment
    experiment2
  6. Give the experiment a name. This is just for your use. Make it meaningful IE. Homepage test 3
    experiments3.gif
  7. Choose the page that you wish to experiment with. Use the link selector. In this case we are using the original homepage
    Choose the new page that you recently duplicated and optimised
  8. Select the goals that you would like to measure the performance of your test
  9. Save your experiment. Make sure that you enable the test to make it live
  10. Let the test run for a period of time. Check your report and interpret the data to determine a conclusive result. It may take some time to collect enough data to ensure that the experiment is conclusive. The duration of your test will depend upon the amount of traffic to the website and the conversion rate for the goals of each page.
  11. Once you have a conclusive result and if your new page is the winner, you will need to implement the changes of your successful page back to the original page. The best way to do this is to rebuild the page by copy and pasting the source code of individual components used.
  12. Repeat the steps all over again to create a new test. Be careful, Split testing is addictive :)

Interpreting split test reports


split-test-report.gif
  1. You can access the report for your split tests at any time by selecting the report link next to each experiment
  2. The trend graph at the top shows the total traffic to the test that view both pages. The blue line represents total visitors and the green line represents the return visitors
  3. The report shows success metrics for both pages in the experiment and compares them side by side. It provides a comparison for each of the goals that you specified for your test
  4. For each goal the report compares total number of goals completed and the conversion rate for each page. For example with our experiment above we are testing the homepage and want to know how many people click through to view a product page. The "Visits that visited any product page" tells us exactly how many visits to the homepage that clicked through to view at least one product page.
  5. The experiment tool will work its magic behind the scenes for you and suggest a winner when you have enough data. A small red rosette will be displayed next to your winning goal for the winning page. You can learn more about the computers choice by selecting the "Winner" or Loser" link next to each conclusive result Learn how the magic works