Prepare an A/B test for websites with low traffic



Business Benefits

Improve your conversion rate.

Choose the most appropriate A/B testing platform for you.

If you haven’t done A/B testing before, consider starting with Google Optimize because it is free and integrated with Google Analytics.

Decide on the question you wish to answer.

For example, you might want to know whether changing a headline on a blog post will increase conversion from a landing page further down the funnel.

Decide on what elements you need to test.

Based on the question you wish to answer, pick an element that needs to be tested. In the example above it would be the headline.

Pick a conversion point near to the element you wish to measure.

The more steps between the element you want to test and the conversion point, the more traffic will leave your funnel. Because you don’t have a lot of traffic that can be a problem.

To avoid this problem, make sure the conversion point is near the element you are testing. For example, if you are testing a blog post headline, do not test whether it leads to a conversion. Instead, test if it encourages people to take the next step in the sales funnel, like clicking on the headline to read the post.

Create a single variation of the element you want to test.

The more variations you have, the less traffic sees each version. Only test one variation of your chosen element at a time with a 50/50 traffic split.

Try creating a bold variation with big changes.

Subtle changes, like changing a headline’s color, will not have a big impact on conversion, making it hard to tell if your variation will make a big difference. Instead, make a big change like changing the size and position of a headline, that will potentially make a bigger improvement and show clearly if your variation will win.

Judge success based on intuition, not statistical thresholds.

Don’t wait for your A/B software to call a winner. With low levels of traffic this will take a long time. Instead, look at your results and make a judgement based on what you see.

If in doubt, supplement A/B testing with qualitative testing.

If you are unsure what judgement to make because results are close, consider doing some qualitative testing such as usability testing.

1 Like