This article will help you:
  • Create an experiment that makes different changes across multiple pages, like a funnel or a site-wide experience (if you're making the same change to multiple pages, see URL Targeting)
  • Set up a multi-page experiment in Optimizely
  • Add pages and variations to a multi-page experiment
  • Set URL Targeting for each page
 
Note:

Multi-Page (Funnel) Experiments are not available to all plan types. If you don’t see this feature in your Optimizely account, or want to learn more about what’s available, please refer to our pricing matrix.

For an overview of the pros and cons of multi-page experiments, see our article on experiment types.

Multi-page experiments allow you to link together variations of different pages. For example, visitors who saw the original version of Page 1 will also see the original version of Page 2, visitors who saw Variation 1 of Page 1 will also see Variation 1 of Page 2, and so on. This is particularly helpful for funnel testing.

 
Note:

You do not need to use a multi-page test just to change an element that shows up on more than one page. For example, if you want to change the look of your top navigation links for every page of your site, you would use a URL targeting condition that targets every page that has the navigation.

Create a multi-page experiment

To create a multi-page experiment, start the same way you would to create a regular experiment: load a page into the Editor. 

  1. Once in the editor, click Options > Experiment Type.


     
  2. Change the setting from A/B Test to Multi-page Test.

Add pages

After you select Multi-page Test, you will see the + Add Page option above the Variation Menus.

This lets you add additional pages to your experiment, so that you can create variations that change multiple elements along a funnel. For example, you could create a multi-page experiment that tests a 50% off promotion and makes changes to:

  • Your landing page
  • Your product pages (see the targeting section below for information on how to target multiple pages at once)
  • Your cart page
  • Your checkout page

In a multi-page experiment, a visitor who sees a variation on any of these pages will continue to see that variation through the rest of the pages that are part of the multi-page experiment.

Add a page by clicking the Add Page button, then entering a name and URL.

You can change the name of the variation by clicking on the arrow next to the variation name and selecting Rename Variation.

 
Note:

Visitors do not need to visit your pages in any particular sequence to be bucketed into a multi-page experiment. Once they visit any pages in your multi-page experiment, they will remain in that variation when visiting any other page in the experiment.

Add variations

Create the variations that you would like to test across your multi-page experiment. The example below shows two variations: "50% OFF" and "BUY ONE GET ONE FREE." These variations should apply to every page in your experiment.

In a multi-page experiment, the pages you add will automatically have the same variation names that were included in the first page. This is because multi-page experiments are intended to test a complete experience across multiple pages. The screenshots below show a single variation across multiple pages.



Remember, with a multi-page experiment visitors who saw the original version of Page 1 will also see the original version of Page 2, visitors who saw Variation 1 of Page 1 will also see Variation 1 of Page 2, and so on.

Once you have finished setting up a multi-page experiment you can add goals, set targeting, and adjust your traffic allocation just as you would in a regular A/B experiment. See below for additional information on targeting.

Set URL Targeting and Audiences

In Optimizely, URL targeting conditions apply to each page in a multi-page experiment. Each page has its own targeting conditions, which means you can apply changes not just to individual pages, but multiple page types.

 
Important:

If a certain URL on your website matches the targeting conditions of two separate “pages” within the multi-page test, the variation code for both of those variations will execute.

Here is an example of when you might want to use a multi-page test for an e-commerce store:

  • Page one of your test is the homepage. 
  • Page two of the test is all category pages.
  • Page three of your test is all product pages.

The goal is to have users who see some experience on the homepage, see a corresponding experience on the other pages. For each “page” type, you will specify targeting conditions. Click Options > URL Targeting and you will see that each “page” that you have set up has its own set of URL targeting conditions.

For example, let’s say we are running a multi-page test on Optimizely’s home page (www.optimizely.com) and our pricing page (www.optimizely.com/pricing). We would not want to use a substring match on www.optimizely.com for the homepage targeting. This is because the pricing page www.optimizely.com/pricing would also pass the targeting conditions for this URL match type.

In this scenario, if a single website URL passed two separate targeting conditions in a multi-page test, variation code from both of these “pages” will run.

 
 
Note:

When you exclude a URL for a page in a multi-page experiment, Optimizely will not run your variation code for the visitor, but visitors will still be bucketed into that variation on the results page. This means that if you exclude a URL, your visitors won’t see any change on those pages. However, their pageviews and conversions will still show up as part of the experiment. Excluded visitors will be randomly distributed between your variations, so these excluded visitors won’t affect your data. However, it may take longer for you to get significant results from a test.

Unlike URL Targeting, Audiences must apply across all pages of the multi-page experiment. In order for a visitor to be bucketed into a multi-page experiment, they must match both the URL targeting conditions and all the audience conditions. In this sense, setting audiences for a multi-page experiment follows the same process as an A/B experiment.