Test landing pages

About landing page tests

When a consumer searches for televisions, where should the ad click go? A promotional page? A product details page? Premium vs. budget product pages?

The DoubleClick Search (DS) landing page testing feature helps you answer this question by testing different landing pages to create the optimal post-click experience for the consumer. Post-click data gives you the information needed to identify pages that produce the best return on investment. You can then make changes to keyword-level landing pages based on these insights. After that, use DS reports to calculate the exact impact your landing page optimizations have made, even months after the tests have been run.

With landing page testing, you can:

  • Split search ad traffic among multiple landing pages and run randomized A/B tests. You can send equal amounts of traffic to all landing pages, or specify the percentage of traffic that each page in the test should receive.

  • Scope the test to specific keywords, or all keywords in an ad group, campaign, or engine account.

  • Report and compare conversions from each landing page in the test. Landing page tests can report on conversions recorded by Floodlight activities or by Google Analytics.

  • See a history of all tests that have been completed.

URLs and landing page tests

Each keyword that you add to a landing page test is associated with the following URLs:

  • The original landing page URL that you specified when you created the keyword. The test does not use or modify this URL.

  • The control landing page URL that you specify when you create the test. Usually, you’ll specify a URL for a landing page that you’ve been using for awhile. DS will redirect to this URL instead of the keyword’s original URL for a percentage of the clicks that a keyword generates.

  • One or more experiment landing page URLs that you specify when you create the test. DS will redirect to these URLs instead of the keyword’s original URL for a percentage of the clicks that a keyword generates.

URL requirements

Note the following requirements for URLs in landing page tests:

Migrating to AdWords Upgraded URLs

New landing page tests can't be started in AdWords engine accounts until you migrate the account to Upgraded URLs.

Original landing page URL

Even though landing page tests do not use the original landing page URL, note the following requirements for original landing page URLs:

  • No dynamic parameters in the main portion of the URL:
    In the keyword’s original landing page URL, the protocol, hostname, and path portion cannot contain dynamic parameters, such as ValueTrack parameters in AdWords or dynamic text in Bing Ads. For example, you cannot test a keyword with any of the following original landing page URLs:

    http://www.example.com/{ifsearch:search}{ifcontent:content}
    http://{ifmobile:m.example.com}{ifnotmobile:www.example.com}

    If you’re using the ifmobile and ifnotmobile parameters to direct traffic to different device types, you'll need to use an alternate form of the device-specific URL.

  • Dynamic parameters in the query string:
    If the original landing page URL contains a dynamic parameter in the query string, DS will add the query string to the test’s clickserver URL. The engine will then provide a value for the query string, and DS appends the query string just before it redirects to the control or experiment URL.
  • Static parameters in the query string:
    If the original landing page URL contains a static query string, the engine's parameter-handling option determines whether DS adds the static query string to the test’s clickserver URL. If you select the Include all parameters (not full URL) option, then DS appends static query strings before it redirects to the control or experiment URL.
For example:
  1. A keyword’s original landing page URL is
    http://www.example.com/snow?matchtype={matchtype}&p_id_static=42

  2. You create a landing page test with a control URL of
    http://www.example.com/snow-test-control
    and an experiment URL of
    http://www.example.com/snow-test-new

  3. You set the engine to add static query strings to clickserver URLs as well as dynamic query strings.

  4. When the keyword generates a click, the engine will provide the value of {matchtype}. DS then redirects to one of the following URLs based on the traffic splitting percentage you specified:
    http://www.example.com/snow-test-control?matchtype=b&p_id_static=42
    http://www.example.com/snow-test-new?matchtype=b&p_id_static=42

Control and experiment URLs

  • No dynamic parameters:
    The control and experiment landing page URLs that you specify in a test cannot contain dynamic parameters. For example, you cannot specify the following for a control URL or experiment URL:
    http://www.example.com/{ifsearch:search}{ifcontent:content}
    http://www.example.com/snow?matchtype={matchtype}
  • Static parameters:
    You can add static URL parameters to the control URL or experiment URL, and these parameters can use DS macros. However, DS does not de-duplicate parameters that are both in the clickserver URL and in the test URLs.

Parameter handling setting for the engine account

Before you set up a landing page test, make sure that the parameter-handling setting in the engine account is not set to Include full URL (unencoded). You cannot run a landing page test with this parameter-handling option.

If you're using the default Include only dynamic parameters option, your test URLs will need to specify any static URL parameters that are required by your site. With this option, DS does not pass static parameters from the original landing page URLs to the test URLs.

If you're using the Include all parameters (not full URL) option, you could end up with duplicate parameters if both your original landing page URLs and your test URLs specify the same parameters.

To keep track of which keywords are in your test, create a label and apply it to the keywords.

How helpful is this article:

Feedback recorded. Thanks!
  • Not at all helpful
  • Not very helpful
  • Somewhat helpful
  • Very helpful
  • Extremely helpful