Target URLs to choose where your experiment runs

Choose the URLs that a visitor needs to use to run an experiment. You also can use the URL match types available in Optimizely Web Experimentation or Optimizely Performance Edge: simple, exact, substring, and regular expression (RegEx).

URL targeting

After you create variations in the editor, determine where the experiment runs on your site. Use the match types to target a specific URL or a group of URLs. Make sure the Optimizely Web Experimentation or Optimizely Performance Edge snippet is implemented on all pages you target.

Find URL targeting

1. Select your experiment from the Experiments page.

2. Click Targeting.

3. Select Target By > URL.

4. Configure URL targeting for your experiment.

5. Click Save.

Test URLs to check targeting

Expand Test URL(s) to check your targeting conditions and ensure that your experiment runs on the expected pages. Test URL(s) determines that an experiment will run if the entered sample URLs match the URL targeting conditions. A non-match means the experiment will not run.

Test URL(s) confirms whether a URL targeting condition or pageview goal correctly targets the URLs you intend. Click Add (+) to add a URL to validate. Try checking for URLs that expect and do not expect to match in your experiment.

URL match types

In an experiment on the Targeting page, enter the URL you want to use and add URL match specifications that tell Optimizely Web Experimentation or Optimizely Performance Edge where the experiment should run.

Each URL match you enter can be one of these types:

• Simple match – This is the default match type. It is useful for testing a single page.

• Exact match – Use only if the URL significantly changes how the page displays due to added query parameters or hash parameters. To target visitors who are using a certain query parameter, use an audience condition instead of an exact match.

• Substring match – Use to match specific strings of text within a URL. It is useful for experiments that change the same element site-wide or on multiple pages.

• Regular expression – Use to target complicated URL structures that are not easily captured by the other URL match types.

Simple match

Simple match is ideal for running experiments on single pages. A simple match runs when visitors land on a URL, even if it has the following variations:

• Add or change a query parameter
  http://www.atticandbutton.us/?utm_campaign=january
• Add or change a hash parameter
  http://www.atticandbutton.us#about
• Change protocol between http:// and https://
  http://www.atticandbutton.us and https://www.atticandbutton.us
• Add or remove a trailing slash at the end of your URL
  http://www.atticandbutton.us and http://www.atticandbutton.us/
• Add or remove www from your URL
  http://www.atticandbutton.us and http://atticandbutton.us

Simple match ignores these changes (because they usually do not change the way your page displays) and runs your experiment on the page.

To use simple match, enter the full URL of the page you want to target—the absolute URL, not a relative URL like /products/general.html. The following examples show common variants that pass a simple match:

These examples match because they run on the same page as the original URL and adding query (?) or hash (#) parameters does not generally change the page.

Simple match will not ignore changes in subdomain (aside from www), adding .html or other extensions, or adding or changing subdirectories.

The following examples will not pass a simple match:

These examples will not match the original URL because they are not the same page:

• https://www.atticandbutton.com does not have the same extension as https://www.atticandbutton.us

• https://www.atticandbutton.us does not have the same subdirectory as https://www.atticandbutton.us/about

• https://www.blog.atticandbutton.us does not have the same subdomain as https://www.atticandbutton.us

Example – User account page with many dynamic/unique query string values

You want to run an experiment on a My Account page for all of your customers. However, every account URL includes query parameters that are unique to a particular user’s information. A sample account URL is:
www.atticandbutton.us/myaccount?account_id=1234&location=san_francisco

To set it up:

1. Identify the static and dynamic portions of the account page URL.

   In this case, the URL to the left of the “?” is always the same (static), and the parameters to the right of the “?” are unique to each user (dynamic).

2. In URL targeting, select Simple Match from the drop-down list. Add the account URL, excluding query string parameters (in this case, www.atticandbutton.us/myaccount).

   A simple match ignores anything to the right of the question mark in the URL (all query string parameters). In this case, ?account_id=1234&location=san_francisco is ignored, which is what you want since this is the dynamic portion of the account page URL.

This setup ensures that your experiment runs on account pages, regardless of the unique user who is viewing the page.

Example – Visitors who land on only two of many landing pages

You have about 10 live landing pages, and you want to run an A/B test on only two of these pages. You know there are unique UTM/query parameters added to the URL from paid search traffic, but you want the experiment to run regardless of which campaign a visitor comes from.

To set it up:

1. Identify the two specific URLs you want to test and how they differ from other landing page URLs on your domain.
2. On the Targeting page of your experiment, select Simple Match from the drop-down list and add the URL of the first landing page.
3. Click Add (+) and add the URL of the second landing page as a Simple Match.

Because a simple match ignores anything to the right of the question mark in the URL (all query string parameters), this setup ensures the experiment runs only if a visitor lands on the first landing page URL or the second landing page URL.

Simple match may also be useful for excluding certain URLs from a substring match that targets many pages at once.

In these examples, those that include www.atticandbutton.us/about are excluded because the targeting is set to exclude that URL as a simple match.

Exact match

Exact match runs an experiment only when users load the exact URL.

Like simple match, exact match ignores trailing slashes at the end of your URL, http and https protocols, and www. Also, like simple match, exact match does not ignore subdirectories, extensions, or changes in subdomains (www.atticandbutton.us compared to blog.atticandbutton.us or www.atticandbutton.us/blog).

Unlike simple match, exact match does not ignore changes to query or hash parameters.

Enter a full URL when using exact match—the absolute URL, not a relative URL like /products/general.html.

In this example, www.atticandbutton.us, http://atticandbutton.us/, and https://www.atticandbutton.us are included because exact match accepts these common variations.

The following URLs are excluded:

• http://www.atticandbutton.us?query=true because it includes a query parameter.
• http://www.atticandbutton.us/cart because it is not the same subdirectory.
• http://blog.atticandbutton.us because it is not the same subdomain.

If you specify a query or hash parameter, exact match allows only that exact parameter into the experiment.

Exact match may be useful when you want to exclude specific query or hash parameters from a simple match.

Substring match

Substring match runs an experiment when the URL contains a set of characters that you specify. This set of characters is called a substring and can be present anywhere in the URL. 

For example, if you want to target an element on all product pages, look for a part of your URL that is consistent across all product pages. If the substring /products is in every product page, run a substring match on /products.

Substring match ignores trailing slashes that display at the end of your URL, and http and https protocols, but it does not ignore subdomains (including www), subdirectories, extensions, or changes to query or hash parameters. To match a substring that includes a trailing slash (if you have a page like /example and need to match /ex/ instead), use a regex match instead of substring.

Unlike simple and exact match, substring match does not need a full URL to match. You can substring-match on a word, query parameter, subdirectory, or any other substring that might display somewhere in the URL. This makes substring match ideal for targeting entire categories of pages or your entire site (for example, to make a change to your global navigation menu).

The following image shows how a substring match targets the experiment when you include www (or another subdomain):

In this example, only the URLs that include the string www.atticandbutton.us are included.

But when you remove the subdomain or www, the substring match targets any page on your domain where the snippet is implemented:

You can target an entire category of pages by including the directory that displays in their URL:

In this example, all strings that include the string accessories are included.

With substring match, you can also target a certain parameter anywhere it displays:

In this example, all the URLs that include the string utm_campaign=springpromo are included.

Example – Testing on categories of pages

You want to run an experiment on all product detail pages, but you want to be sure that your changes do not also show up on other pages of the site. You know that all product detail page URLs look similar (for example, http://www.atticandbutton.us/product/scarf?p=4673, where everything after product is dynamic).

To set it up:

1. Identify a common URL structure for all of your product pages that is unique to those pages.
2. On the Targeting page of your experiment, select Substring Match from the drop-down list and enter the portion of the URL that is guaranteed to be the same across the desired set of pages (for example, atticandbutton.us/product).

This ensures that the experiment runs on any URL that has the string atticandbutton.us/product in the full URL. The substring match applies even if values are added before or after this string.

Example – Run site-wide, except for a certain subset of pages

You want to run a site-wide navigation bar change across all the main pages on your site. However, a subset of news pages do not follow your site’s standard layout, so you want to exclude these news pages from the experiment.

You know that the news pages all have a URL that begins with www.atticandbutton.us/news. The dynamic topic and article name are appended after news/.

For example, a full URL might be: www.atticandbutton.us/news/home-improvement-projects

To set it up:

1. Start with the bigger group of pages that you want to include in the experiment. In this case, set a substring match to www.atticandbutton.us to ensure the experiment runs site-wide.
2. Identify a common URL structure for all of the pages that you want to exclude from the experiment.
3. On the Targeting page of your experiment, add another condition and select Substring Match from the drop-down list. Enter the portion of the URL that is guaranteed to be the same across the set of pages you want to exclude (in this example, www.atticandbutton.us/news works).

The experiment runs on pages on your domain that have the substring www.atticandbutton.us, but not pages that have the string www.atticandbutton.us/news.

When you exclude URLs in URL targeting, both positive and negative conditions must be met. In other words, to be included in the experiment, a visitor must land on a page that matches the included condition and does not match the excluded condition.

Regular expression match

Expert users can use regular expressions (RegEx) to specify a set of pages on which an experiment should run. For example, to target pages on your site where the snippet is implemented, use a regular expression match and enter .* as your match condition.

You cannot cover all of the possibilities that regular expressions allow. For example, suppose that you want to target your experiment to pages in your /products/shoes, /products/hats, and /products/shirts sections, but no others. You might write a regular expression that looks like this: /products/(shoes|hats|shirts).*.

Regular expression matches are evaluated using JavaScript's built-in regular RegEx module. Your expression is matched against the full page URL and is evaluated as a substring match up until an escaped character in your expression. Alternatively, you could write the regular expression in this example entirely in RegEx: \/products\/shoes|hats|shirts.*.

Your expression does not need to match the full URL to trigger your experiment. For example, the regular expression x=y will match the following pages:

• http://www.atticandbutton.us/blog/january?x=y
• https://atticandbutton.us/blog/january?a=b&x=y&c=d

Query parameters

When Optimizely evaluates a URL using the RegEx match type, Optimizely strips the Optimizely Experimentation query parameters. For example, if you are trying to create a RegEx to match http://www.atticandbutton.us, the added Optimizely Experimentation query parameters might look like this:

http://www.atticandbutton.us/? optimizely_token=b0a1ff184499db5eba13276b436f7a9c04422c85ed08ac7cb83b8214768b37be& optimizely_x=7961690747& optimizely_x_audiences=& optimizely_preview_layer_ids=7597420149& optimizely_snippet=s3-7428110562& optimizely_preview_mode_CAMPAIGN=7597420149.

The stripped URL Optimizely evaluates would be http://www.atticandbutton.us/?. Optimizely does not remove the /? when stripping the Optimizely Experimentation query parameters due to inconsistent behavior among browsers.

To match the URL http://www.atticandbutton.us, you may want to include matching for the trailing /?. You could use example\.com\/?\??$ to match with and without the trailing forward slash and question mark and to not match if there is an actual query parameter in the URL, like a UTM campaign.

Resources

• To control who can see the experiment, use audiences.
• To edit a site-wide element, such as a button in the navigation bar or a CTA in the footer, see Edit elements that display site-wide.
• Have a question about targeting? See the Optimizely Experimentation questions on Stack Overflow to join the discussion.