Grow Your PPC Revenue – AdWords Features and Experiments You Should Consider
Will Fu May 24 2018
As a PPC Strategist working with a variety of clients and industries, I see often certain AdWords features work for some and not others. While there are some obvious wins or constant sources of high ROI, such as Remarketing Lists for Search Ads (RLSAs) producing better revenue and user engagement than broad targeting Display ads, it’s important to try all of AdWords features to find the best efficiencies.
In this post, we’ll share some of the awesome features AdWords has available and encourage you to experiment with any or all of these to see what works for your growing business.
Important: These features are not an alternative to building out the core of your SEM campaign using your deep understanding of real prospects and customers. There is no substitute for great topic research, and keyword research, driven by customer journey mapping, and a clear picture of what your customers care about.
A fool with a tool is still a fool.
– Grady Booch
Smart Display campaigns are the “pinnacle of programmatic” as Google keeps telling us. And in terms of giving up control to Google, they may be right. These campaigns are for the Google Display Network (GDN) and are highly automated.
All you provide as the advertiser are the ideal cost per conversion (Target CPA Bid), campaign budget, and landing page or creative assets.
In exchange, Google provides auto-optimized creatives, the “right” users with the “right” bid, and coverage across the entire GDN inventory. That’s right, no more creating placements, finding audiences, and bid adjustments.
In general, this experiment will perform the best for AdWords accounts that have:
- A clear target cost-per-acquisition (CPA)
- A short conversion lag
- 100+ Search or 50+ Display conversions per month
- A remarketing tag
Dynamic Search Ads
Dynamic Search Ads (DSA) are similar to Smart Display campaigns in that there’s a lot of automation straight out of the box. But instead of the Google Display Network, DSAs are shown on the Google Search Network, so this campaign type is listed under Search Network Only campaigns.
Broadly, Google uses information from your site, compares that information with search queries of your site visitors, and then creates ads (both headlines and landing pages) to best match the users it targets on your behalf. If you have concerns about how well your keywords, ad copy, and Final URL match up, this is the experiment for you.
DSAs can be set up quickly and require minimal hands-on optimization. Here’s what you’ll need to provide:
What landing pages should ads link to? (There are three options.)
- Target categories where you select sites based on ad groups
- Target all pages
- Target specific pages
Ad description lines and an optional display URL
- The headline and final URL will be dynamically generated
Other optimizations similar to standard Search campaigns
- Ad extensions
- Ad schedule
- Bid adjustments
- Negative keywords
There are an absolutely incredible variety of possibilities for retargeting (aka remarketing). Create an audience for customers who visited your site, clicked a link, stayed on your site for 10 seconds, 30 seconds, abandoned the checkout process, performed actions in an exact order, and multi-factored audiences. If you’re not already well on your way with retargeting, test all of these individually (using intuition to guide what makes the best sense for your buyer’s journey) to find the ideal target audience for both search and display networks.
But don’t call it good there! Once you have a great target remarketing audience, look to expand on it immediately with similar audiences or tailor that remarketing audience first to guide what would go into similar audience. Again, look to do this for both Search and Display networks since the medium will change your strategy and messaging by necessity.
Pro Tip: Note that similar audiences do have some requirements. Similar audiences for Search require 1,000 cookies with enough similarity in the original audience.
This kind of experiment or campaign can seem overwhelming but is absolutely worth testing thoroughly. For example, if you have two initial remarketing audiences based on observed behaviors on your site (say specific page views and length of time on site), and you’re advertising in both Search and Display, you’ll have eight audiences to test:
- Remarketing audience 1 for Search
- Remarketing audience 1 for Display
- Remarketing audience 2 for Search
- Remarketing audience 2 for Display
- Similar to remarketing audience 1 for Search
- Similar to remarketing audience 1 for Display
- Similar to remarketing audience 2 for Search
- Similar to remarketing audience 2 for Display
Pro-Tip: Tack on the Google-defined audiences of Affinity, In-Market, Life events, and Custom audiences for Search or Display campaigns and test the results with any applicable audience.
Last is customer match where you can upload customer information and retarget them through Search, Gmail Sponsored Promotions, or Youtube ads. Here’s a little more on how to set up customer match. Customer information ranges from email, phone number, mailing address, and advertising IDs.
Smart Bidding Strategies
If you’re on manual CPC, don’t have a lot of time to spend in AdWords, or are unsure of how to optimize, then this may be the feature experiment for you.
Smart Bidding Strategies such as Target CPA, maximize conversions, or Target ROAS gives AdWords a lot more control over optimizations and usually does a great job. Like any other AdWords feature, test to make sure it works for your account and industry. Check out Google’s knowledge base article to learn more about the goals you can use to optimize with this machine learning approach.
The main reason to switch bidding strategies is the relative difficulty and time required to keep on top of bid adjustments when you’re managing it yourself. You’re absolutely giving up some control as with many of the other features here, but you’re able to scale up campaigns faster, which may be the right trade-off for some.
Multiply the combinations together and you get 1,764 different permutations to manage for a single campaign. Add on location and ad scheduling by the hour and this can quickly get out of hand.
Experiment with Attribution Models in AdWords
For Portent, and as it becomes increasingly available to more advertisers, it’s all about Google’s recommended “Data-driven” attribution model. (And yes, “Data-driven” is actually what they’ve named this ML approach to attribution.) Data-driven does require some scale at 15,000+ clicks and 600+ conversions in a month. If you’re coming up short of this, but don’t want the default last-click attribution that GA offers (good for you), the options for Linear attribution or Position-Based attribution do not.
Picking one of these 3 attribution models and using it to optimize performance often improves bottom-line KPIs whether it’s conversion quality, ROAS, or GPM by causation or LTV and retention by correlation. Again, that’s because AdWords’ default attribution model is last-click, which means whatever the user clicks last when he or she converts gets 100% of the conversion credit, which can lead to some serious problems awarding credit to other important touch-points.
Think about how you are reaching customers and pick which attribution model is best for you. You can compare attribution models in your AdWords account. Tools => Conversions => (pick a conversion) => Edit Settings => (click Attribution model to expand) => click “attribution modeling tool.”
(BONUS) A Few More Feature-Based Experiments to Try in AdWords
Custom Columns – These are great for GPM and PM, but in truth these are more of a reporting feature than something to drive or automate the performance of the campaigns themselves.
Beta & Alpha testing programs – Contact your Google Representative to apply for Beta or Alpha programs that give additional targeting. You may be surprised at who qualifies. For example, the new Detailed Demographics gives advertisers the ability to target users by Marital Status, Home Ownership Status, Education Level, and Parental Status. We’ll wait and see how this one plays out with GDPR.