We are excited to announce multi-page experiments on Optimizely! Lots of our customers want to A/B test not just one, but two or more pages that are related, often as part of a funnel. Multi-page experiments allow you to “link” together related variations of different pages: visitors who see the original version of Page 1 will also see the the original version of Page 2, visitors who saw a specific variation of Page 1, will see a corresponding variation of Page 2. This allows you to maintain a consistent user experience through a funnel or even across your whole website. Multi-page experiments are available to Optimizely customers with Gold and Platinum plans.
As an example, let’s say we want to test whether customers are more likely to buy an item called a “large coffee mug” or a “portable coffee reservoir.” If we ran an A/B test on the product page only, visitors who saw the “Reservoir” variation would still see the “Original” version of the checkout page and might think the wrong product was added to the cart:
With a multi-page experiment, we can create “linked” variations across both of these pages to provide consistent product marketing. Start by selecting the “Multi-page” experiment type:
Now we can add the checkout page to our experiment. Notice that a “Reservoir” variation, linked to the product page “Reservoir” variation, will automatically appear:
We can then modify the “Reservoir” variation of the checkout page to be consistent with the “Reservoir” variation of the product page. Here, we will change the product name to be consistent with the product page “Reservoir” variation:
Variations with the same name, including the “Original” (or control) versions, will be “linked” together ensuring a consistent user experience across the pages included in the experiment. In fact, if this same product was referred to as a “large coffee mug” on a third page, we could go ahead and add that page to the multi-page experiment as well.
The pages in a multi-page experiment do not need to be viewed in any particular order, or even at all, by a visitor.