How to configure targeting of experiments with Optimizely vs Google Optimize

This article is comparing targeting capabilities of where tests run available in Optimizely vs Google Optimize.

Importance of targeting for A/B tests

Before you can start running an A/B test you need to set up targeting of where the experiment should run on your website. This can be either one page, a group of pages or all pages of the website.?

Let’s take a look at how targeting can be set up in both tools.

Targeting with Google Optimize?

Google Optimize allows to specify page targeting by URL as a single trigger with limited matching rules.?

For example, a test can run on one page where the URL contains a certain fragment such as ‘/product-details/’. If you want to target more specific pages then you can use regular expressions. It works well for dynamic websites where you can determine a segment of pages with a regular expression.?

One of requirements in my most recent project was to target all ‘coming soon’ product pages. This was achieved by concatenating product IDs into a single regular expression string e.g. ‘productID1|productID2|productID3…’ where a developer exported a list of all relevant product IDs form the database using a PowerShell script, and had to do it on a regular basis.?

The screenshot below illustrates the only targeting options in Google Optimize.?

The major concern with this approach is that you can’t edit a regular expression in a running test. The only option for frequently changing product catalog websites is to keep removing previously configured regular expression and replacing it with an updated one.?

In my experience I had to maintain a spreadsheet of product IDs with a formula to concatenate them into a single string regular expression. This was a manual task which had to be done repeatedly, less than ideal. The screenshot below shows an example of an incremental concatenation approach where the last string is my final value of a regular expression.

Targeting with Optimizely Web Experimentation?

Unlike Google Optimize, Optimizely offers more flexible options for targeting of where tests can run.?

Classic approach with page URL is also available with added flexibility to meet more complex requirements. If you want to target a specific segment of pages you can use regular expressions which can be edited at any time.?

You can also create templates containing complex targeting rules and reuse them in multiple tests. This will save you time on targeting configuration and future maintenance of tests.

In addition to classic URL match conditions, Optimizely offers two more advanced targeting options:

• Element is present - allows to check if an element is present on the page. For example, I used this option to run a test on all pages containing product listings which I targeted using a CSS selector which can only be present in product listing component HTML.?
• JavaScirpot condition - supports writing your own conditional logic for specific use cases. For example, I used this option to run a test on all pages containing product listings with previously viewed products.?

You can even construct combinations of complex rules from all three of the above conditions, each condition type can be used multiple times.?

A/B testing with Single Page Applications (SPAs)

Google Optimize can only be used on typical websites where every page has its own URL. It was never designed to be used in single page applications where the URL remains unchanged while users navigate through different pages.

Optimizely Web Experimentation has a solution for single page applications.

For example, I can use it on a website where a bag page and the rest of the checkout pages open in a pop up while the URL remains as the product details page URL.?

Optimizely tool can handle such tricky use cases. It makes it possible to run tests on SPA websites with its more advanced targeting conditions.

Summary

Whilst Google Optimize and Optimizely have some similarities for targeting where tests should run, provision of flexible configuration for modern web applications in Optimizely is on a different level. Google Optimize offers only basic features which are not enough for more sophisticated experimentation programs.??

Google Optimize is sunsetting in September, but it not a problem for brads with more complex website requirements because this tool was never suitable for advanced testing practices such as testing on single page applications or ease of maintenance for concurrently running tests and their targeting.?

Optimizely Web Experimentation tool is a great alternative to consider.?

In my next post I will talk about another area of where the difference between Optimizely and Google Optimize is significant - setting multiple macro and micro objectives for experiments.?