Dynamic search ads (DSA) are an unusual ad format as you don’t write a headline or pick a destination URL. With these ads, you choose what part of your site you want Google to use to scan for potential keywords and Google will show your ad and the relevant destination URL when the search query matches your landing page. Google also writes the headline for you.
However, you can still test these ad types since you write the description lines and display URL and you can choose what parts of your site to crawl for each ad group.
With DSAs, the first test you should run is determining how granular you should segment your site. With each ‘segment’ you can write description lines and display URLs that are more relevant to the sections of your site that you are using to create DSAs. While you could write just one ad and use it to advertise your entire site, the description lines will be very generic (like all appliances available, free shipping on orders over $150, largest selection available, etc).
As with any ads, the most specific that ad is to the search query, the higher the CTR and conversion rates usually are.
For instance, an appliance retailer approached me asking how to run DSAs across their site. They sell appliances for any room in a house and their appliances usually run 4-8 categories deep per room.
In the kitchen, you can have this category drill down:
Without duplicating the entire structure, each time you add a new category down, your ads explode exponentially. For instance, at the top level you have one ad ‘appliances’. Then you have an ad by room, then you have an ad by small & large, then you have an ad by type (blender), then you have an ad by subtype (espresso maker), etc.
This leads to the possibility of having one DSA or thousands of DSAs – which almost duplicates your existing ad group structure so you really lose the benefit of DSAs which is ‘backfill’ of keywords you don’t have yet.
So the question is: how far should you segment your DSAs before you should stop?
This is a simple test. Pick a few categories, create a very granular DSA structure in just those categories and test to see where adding a new level of DSAs no longer makes a difference in your metrics.
For example, in the test for blenders and juicers, you can see at the level of ‘juicer’ there was no longer an advantage of continuing to segment the site as the impressions dropped significantly and the conversion rate, CTR, and CPI no longer drastically improved at that segment.
Now, a test like this is only going to see a difference in your stats if you create ads that talk to that segment, such as ‘large selection of blenders’ instead of ‘large selection of kitchen appliances’.
This first test will show you how far to segment the site.
The next text will help you determine what ads are best to use across one or more DSAs.
When we consider a typical text ad; there are two ways they can be tested:
As DSAs cover multiple pages of your site, they are often tested within an ad group (you create 2 or more DSAs within an ad group); however, how they are tested is more akin to how you would think of a multi-ad group test.
With a multi-ad group test, you are often trying to see what overall concept works best:
What matters more in your niche? Uniqueness, free shipping, large selection, exclusive, customer service, or something else?
With multi-ad group tests, you are gaining an insight or a line that you can use in multiple ad groups at once. As each ad group is different, this is a global insight that you can use across all of those ad groups.
Since DSAs also cover multiple pages of your site, this is akin to finding ad lines that work across multiple ad groups.
So when you test DSAs, and you should, you’ll want to think about your multi-ad group tests or your overall value propositions that work for multiple ad groups and use those in your testing. Then all you need to do is create two or more DSAs and have each DSA utilize one of your main value propositions so that you can see which DSA is best for that segment of your website.
Multi-ad group tests are always good to have running in the background even if you are using mostly single ad group tests as they can keep you apprised of global insights and trends in your account as to how users are behaving.
When you are testing DSAs, you are testing messages that work across multiple ad groups. It is a good idea to setup those same tests in your typical text ad ad groups as well just to see how those lines are performing across your other campaigns.
This can help you see when you’ve come across a new line for DSAs that will work well in other ad groups; or if your DSA lines aren’t working as well as something else you are using in your other ad groups and you can use those multi-ad group test results to refine and test your DSAs.
DSAs should rarely be your primary source of traffic. While you can pick what parts of your site are used in crawling for keywords, use negative keywords or even negative site phrases (such as out of stock), they are generally best when used as backfill to your hand-picked keywords.
They do make great backfill to increase your overall search query coverage.
However, when we look at accounts – DSAs are rarely tested. Many users think that since they don’t write the headlines or choose the landing pages that DSAs aren’t worth testing. In reality, since DSAs are covering large parts of your site with a single ad; testing them is very important as a slight change in their metrics can increase your conversions quite a bit.
The overall trick to DSAs is to think in terms of writing them like you would a multi-ad group test, but use single ad group testing methods to determine their effectiveness. This is true of most accounts; however, we do have some clients that are managing millions of ads and often tens of thousands of DSAs.
In these larger accounts, you can use multi-ad group testing for DSAs as well. This lets you test at a macro level across your account. These account sizes are the exceptions; so for most people running a few thousand ads – test DSAs within ad groups as a single ad group test but utilize ad lines that work well across your brand or segment of the site you are using for the DSAs.
By testing DSAs in this fashion, you should be able to improve the conversions that your DSAs bring to your account and make better use of this ad format.