By: Ben Sporn
“The two most powerful warriors are patience and time.” – Leo Tolstoy
The Google Display Network (GDN) is one of the largest advertising networks available to marketers and can reach about 90% of all internet users. It has a wide reach and various forms of programmatic targeting and creative ad units. Typically, advertisers struggle to gain traction on GDN because it is such a vast network and can take a lot of pain tolerance through the learning process. Since the level of intent from a user clicking on a display ad is typically lower than most networks, it can take a while to truly find the right users. GDN is all about advertising arbitrage, meaning one can still find positive ROI because the cost-per-click is cheaper, but this must be balanced with lower conversion rates.
**Targeting: **GDN has multiple forms of targeting which we breakdown into three main buckets:
- Audience: Audience targeting is finding users based on their interests. This is calculated based on the types of sites users have visited and the content on those sites users are viewing. For instance, a user who visits CNN will fall into an audience for “News Enthusiasts”. If they are reading content on CNN about tennis, they will also be placed into an audience for “Tennis Enthusiasts”. There are multiple levels of intent with audiences, here are the Google defined names:- - In-Market Audiences (Google Defined): Users who have visited sites recently and frequently related to specific interests.
- Affinity Audience (Google Defined): Users who have visited sites but less frequently and less recently. These are typically your biggest audiences and usually have the lowest conversion rates.
- Keyword Audiences (Custom): This is the audience we typically find the best results with. It is a custom audience based on keywords we enter and we have more control over the sub segments of an audiences. For instance, there may be a In-Market Audience for “Insurance” but to find dental insurance users, we would need to define a set of custom keywords.
- Others: There are other audiences, but the ones above are the most important. You can always target and audience based on demographics for instance (35-45 Females etc.)
- Remarketing: Showing ads back to users who have visited your site.
- Contextual: Contextual audiences help you find users based on the particular content they are presently viewing.
- Display Select Keywords (Custom): This allows you target users when they are presently viewing content related to the keywords you have entered. For instance, a keyword with “Dental Insurance” will find any user who is on a page with that word (other exclusions for demo, device can apply).
- Topics (Google Defined): Topics are a set of Google categories that you can select to target a lot of sites at once. For instance, a topic for financial news will show your ads on sites that fall into that category.
- Placements (Custom): Placements are sites that you choose to show ads on.
- Auto Targeted: Auto targets are categorized as its own category because there is no way to know exactly who you are hitting but multiple ways to leverage auto targets:
- Similar Audiences: Similar audiences is basically a lookalike audience from a seed list. It finds users who look like your converters. The difference between Google’s and Facebook’s lookalike audiences is that Facebook’s can be built from a list of emails which are linked to Facebook accounts. Since the GDN does not have users signed in (you are targeting users on other websites), we cannot link effectively users to a particular Google account.
- Smart Display Campaigns: Smart display campaigns take all of your account data and try to find converters. It is limited with any exclusions so make sure you have a ton of data before turning it on.
- Conservative Auto Targeting: This is a type of targeting that expands the reach of a current audience. It expands the targeting to others users outside of your selected audiences.
**Ads: **GDN ads come in multiple shapes and sizes (simplistic overview)
- Text Ads: These are text only creative that appear on partner sites
- Image Ads (Static): These are manually uploaded images
- Responsive Ads: These are a mix text and image combinations. You have little control over these so you better make sure you know your winning copy/headlines before leveraging them
Bidding: There are multiple ways to bid. Here are a few to test out depending on your budget and goals:
- Manual CPC: You bid based on a click
- eCPC: You bid based on a CPC, but the system can bid up or down if i thinks it can get a conversion
- tCPA: You bid based on the conversions your likely to get
- tROAS: You bid based on the conversion value
How we test: This has been our process for starting clients on the GDN
- We start with audience campaigns segmented by demo and device and bidding on manual CPC. We always start campaigns out on manual CPC because we do not have enough learnings for the system to effectively predict if it will get a conversion.
- The main driver of early performance and a key lever for success is creative testing. We rotate in new creative every 5-7 days and prune out the lowest performing ads. Sometimes we find a headlines that works and we’ll expand it to a different descriptions (or vice versa). Combinations of ad headlines and descriptions are also an important factor and sometimes switching them can lead to higher click-through rates. We are mainly evaluating the performance based on Click Thru Rate and conversion rate (although conversion rate can be misleading with low volume)
- We always layer observable targets on top of GDN targeting for optimizations later on.
- After we find winning audiences, we expand to new keywords or unlock demo targets that we previously gated off.
- Next, we launch contextual targets with Topics and Display select keywords. We are able to jumpstart the learning process for ads because we already have winning copy.
- Next, once we find winning copy we launch responsive ads based on the winning headlines and descriptions.
- After another 5-7 days (or once we reached approximately 250 account conversions) we turn on more automated targeting and bidding such as Smart Display campaigns and tCPA