In this guide, we will cover the following:



1. What is multi-adgroup pattern testing?
Multi-adgroup pattern testing is aimed at split testing sets of ads (across multiple adgroups/campaigns). Each set of ads shares a common text (pattern) in headline1 or headline 2, etc.  The pattern is matched in the same location within all ads.


Multi-adgroup pattern testing helps give you an idea of how well certain messages perform.


Multi-adgroup testing can also be very useful for low volume accounts that don’t accrue enough data to reach statistical significance using single-adgroup testing. 



2. How do I run multi-adgroup pattern tests?

Multi-adgroup test results are produced in 2 ways:


2.1 Manually (run by you)

Just like single-adgroup testing, you can run multi-adgroup tests as often as you need as shown below:




1) Select which campaigns (and optionally specific adgroup labels and ad types) as well as the date range for this test.  A common date range means Adalysis will choose a date range during which all ads within all sets were active.


2) Specify the location within the ad where the pattern will be matched.  One location applies to the whole test.


3) Adalysis analyzes the ad text used by each campaign and, in some instances, will suggest potentially competing patterns (that share some common text) which can provide insights into winning patterns.


4)  Provide the patterns to be tested.  Patterns can be whole text, or can use wildcards to specify more generic text (e.g. Find a plumber in *).  (tip: click on the folder icon to see the top common patterns currently across your chosen set of campaigns in the ad location you specified).


5) Click on this button to run this test.  


6) If this is a test you want Adalysis to run for you daily, click this button to save this test definition in  the Automated Daily Tests screen.

 



Tip 1:  Care needs to be taken when using a common test date range. If one matching ad out of hundreds was enabled only few days ago, it will cause the whole test to run using only the last few days of data. 

Tip 2: Use the Run a sample multi-adgroup test button to see a sample test result from one of your campaign. This sample data is here to help you quickly familiarize yourself with this type of ad testing.

wRWWKaQe1ROkdyonUGU5XG0mhoVUrlQJdA.png




2.2 Automatically (run by Adalysis daily)

Unlike single-adgroup testing, you will first need to define a test before Adalysis can run it daily for you.


Once you run some tests manually (as described above in 2.1) and found one you want to keep running daily, you can do so by using the Save & Run this test daily option in the Run Tests screen or define the test here as shown below:






3. Understanding the test result data

 

1) The result of your last test run

2) Details of each pattern you provided and how many adgroups and ads it was found in.  You can see the individual ads found for each of the patterns. 



P3enoIIh5zb1Xd74Jz21xqFjXdK-5Ckp6Q.png



Once you open the test result details, you will see the aggregate performance of each pattern (over date range used) and the confidence the algorithm has in each of the metrics.  The performance boost figures are projected based on all ads with the losing pattern getting paused


nw5UZY5zjfJySLxX5D8LlakcfQ2d4iJVCw.png




4. What actions can I take when analyzing a test result

Tip: Whenever losing ads are paused, a copy of the test result will be automatically archived for future reference.



4.1 Pausing the loser ads

This will pause all ads that contain the losing pattern for a specific metric.


xC0TF6oacALjn94gFgtq1IcxUZSV6clWzA.png


4.2 Pausing and replacing the loser ads

This will pause all losing ads as well as automatically create a new replacement ad for each losing ad. The replacement ad is identical to the old ad but with the losing pattern replaced by the newly specified pattern.


yJaNbGetbHISU7NFF9rsX9wTwNbquloC4A.png


Tip 1: Use the Preview new Ads option to see what your new ads will look like before you proceed with the creation.

Tip 2:  You must not use wildcards in the new replacement patterns.  You need to only provide the text that will replace the other part of your original pattern.  Example
  1. If the original pattern you're testing includes a wildcard e.g. Find a plumber in *
  2. This will match ads that use the text Find a plumber in FloridaFind a plumber in LA, etc.
  3. If this pattern lost in the test, you can provide a new pattern to replace only the "Find a plumber in" part of the losing pattern.
  4. If you, as an example, provide a new replacement pattern of "Call a plumber in", then the new ads will be created using the full text Call a plumber in Florida, Call a plumber in LA, etc.  i.e. the wildcard part of the old ad is preserved.



5. Viewing historical test results 

 

History of Multi-Adgroup Test Result Changes



6. How pattern matching works in Multi-adgroup Testing


Pattern How Ads are matched to the pattern
abc abc exists exactly as is (with no additional text).
* abc abc exists at the end of the text.
abc * abc exists at the beginning of the text.
abc * xyz abc exists at beginning and xyz at end, with any text in between.
* abc * xyz abc exists anywhere before xyz. xyz exists at end.
abc * xyz * abc exists at beginning. xyz anywhere after abc.
 !!abc  abc doesn't exist as is (or use a wildcard * before/after abc)



7. Related multi-adgroup testing information