Dynamic Search Ads: A Slow and Painful Journey
Elizabeth Marsten Jan 10 2013
Armed once again with intern Tim and a few dollars from Portent’s pocketbook, I decided to take Dynamic Search Ads (DSAs) for a spin using our small business PPC Essentials management package as a test case. I made up my mind pretty quickly and an informal poll of my massive follower list (massive being a relative number) showed the Twitterverse agreed. The consensus was:
DSAs are “more trouble to maintain and ‘optimize’ than they’re worth.”
Let me explain why.
First off, DSAs have a laundry list of contingencies that are enough to make any PPC strategist who’s worked in an SEO agency environment (read: seen/heard a lot about SEO problems over the years) cringe. For example:
- DSAs generate the headline dynamically based on page content
- DSAs generate the landing page dynamically based on crawl index
- Google specifically states that sites that sell customizable products, comparison shopping, daily deals and “certain small websites” will most likely not see a benefit from DSAs.
Guh. The amount of times a developer has said they can’t fix something, claimed they knew SEO or were just “too busy” to edit that robots.txt file that was blocking googlebot from seeing half the site is just the tip of the iceberg on how messy this can get.
Begin the DSA test
Using Portent as my guinea pig once again (and as a result, I can show all my wonderful data), I tried DSAs, refining a little bit each time to see how much I would have to change to the auto targets, how often and the kind of impact the changes might bring. Portent isn’t “perfect” SEO-wise, but it’s far from a hot mess. We’re fast, 10,000-ish pages, good robots.txt, sitemaps, keyword rich this and that, solid navigation, you get the idea.
My first DSA campaign auto target was to focus on “ppc” and exclude “seo.” Because DSAs can’t tell you what categories are available to choose from until you start running (it’s literally a blank slate), you might have to spend a few dollars to find out how Google thinks your site is organized.
When I added a second campaign with a different extension for a sister site (www.analytics-review.com), the category field auto-populated with portent.com suggestions instead, for example.
What I did not try, but would recommend, is the “create and wait” method. Start a DSA campaign and DO NOT enter any categories or targets. Then let it “stew” to see if your category list will populate with suggestions after a couple of days.
Back to our original DSA story… after adding the auto-target of “page_content contains ppc” and excluding “page_title contains seo,” here’s what it my category list looked like after 5 days:
So what search queries did Google show my ads for based on my category of “contains PPC” target?
Ick. And ick. But this is science. Now that I knew what Google thought was a category and what terms it considered relevant to that category, I refined my list further. I added exclusions to the keyword list (april fools, glue gun, jokes, to name a few) and in the URL and page_title areas because the sample of pages (see below) that were being shown with my ads were largely PPC-related but really, not what I had in mind.
The ad copy
You can see the headline and landing page that Google “awarded” me based on the initial criteria I set. Here is the body copy that I created to run with DSAs for the duration of the experiment:
I thought this was pretty straightforward. Nothing about glue guns or pranks here.
DSAs: new and improved!
I feel like I’m fighting a losing battle at this point. For example, here’s what the category list looked like after 20 days:
More Google-suggested categories! Relevant? Not so much. As flattered as I am, I don’t consider myself to be a category. And social media and tools are now showing as categories. I’m not really sure how they ended up that way because the website coverage metric is showing me 0% of pages are indexed. To further refine the auto targets, I added several of these newfound categories as exclusions.
This process is far from intuitive and I am guessing a lot, so I’m not sure I’m getting the best possible results.
The DSA Conclusion
Forty-two days after starting the test, we’ve spent $27.20 on DSAs, had one conversion (which turned out to be a dead lead) 74 clicks, 18,467 impressions and a .40% CTR at .37CPC. While I like the impression rate and CPC, that CTR says that Google didn’t do a great job at choosing my headline.
In all that time I spent trying to teach Google how to write a great headline for me, I most certainly could have written a better one myself.
Here’s the headline/landing page list after 40 days with 3 sets of refinements in negative keywords, negative categories, URLs and page titles:
I circled some of the more amusing items in red. The blue circles represent actual categories of “PPC.” What I might have done differently is not use the auto target of “page_content contains PPC” and instead go more granular with “URL contains.” I could have also changed to the category of PPC once it became available, but I was curious to see what would happen if Google was left to its own devices.
I stand by my earlier statement; DSAs are more trouble than they are worth. I’m sure there are some great case studies out there from people who are loving them. But the poor click-through rate combined with the upfront cost of finding out what Google thinks is relevant and the tweaking thereafter make for a negative ROI. DSAs are a “meh” experience for me.