There are 8 different ways to target ads running on the Google Display Network (GDN). Each of them is based on either the content (keywords) of websites across the GDN, or the behavior of traffic that transits the GDN.
To plan the targeting for a display campaign, we first determine which of the 8 types of targeting are likely to be successful:
For every PPC account and display ad campaign, we create a conversion value model for each type of conversion event and (related), select a method for click attribution modeling. When a conversion occurs, the value for that conversion is distributed ("attributed") up through the sales funnel, with weighting as defined by the click attribution model.
In our conversion value model, what we are after is the gross margin (value) that each conversion action is producing. In the case of a display campaign that advertises products, the conversion value would likely be (sales price-sales tax)*(average gross margin).
Conversion value modeling is important because it allows for calculating the two most important metrics in paid search: return on ad spend, and advertising margin. But more importantly, it provides valuable information back to the automated bidding algorithms, which use machine learning to maximize the performance of the campaign.
Display campaigns are more likely to generate "conversion assist" clicks than final-click conversions. For this reason, click attribution is important for assessing the display campaign's role in conversion performance. But accurate conversion tracking is also important with display.
For this reason, we implement 100% conversion tracking coverage, whether the events being tracked are transactions, form submissions, phone calls, or other meaningful events.
We've found that less than half of existing campaigns have full conversion tracking implemented, and even when conversion tracking is implemented, it is more often than not lacking full coverage.
There is a reason for this – conversion tracking can be difficult to implement and test.
What Makes Display Campaign Optimization ...
After a display campaign launch, they will often burst in an erratic serving pattern during the first 2 or 3 days after launch. This varies by ad platform and campaign design. Average-size display campaigns usually developing a consistent serving pattern within a few days of launch.
Once a display campaign stabilizes, we look for major issues such as a single targeting method or ad group consuming excessive ad spend. We monitor the audiences that pull the campaign into auctions. Typically we make a series of corrective adjustments across our display campaign post-launch checklists, dialing the campaign into the type of serving pattern, or "orbit" that we’re looking for.
When setting up display campaign serving parameters, our preference is usually to allow the campaigns to initially run 24x7x365. We can then use that data to optimize the campaign schedule, as we have measured what times and days of the week are most effective.
In some b2b markets, we will run campaigns on weekdays only and perhaps not overnight. There are some common mistakes to avoid, for example in the US making sure East coast campaigns stay on late enough to shop up for West coast end-of-day business traffic.
The reason for running campaigns (at least initially) 24x7 is that many campaigns perform well at surprising times – days of the week, and hours of the day you would not expect.
In our display campaign competitive assessment process, we familiarize ourselves with how competitors' products are positioned, and the competitive environment the campaign will operate in.
We examine your competitor's advertising history, ad copy, ad spend, and market positioning. We use tools to examine their display advertising history.
The display campaign competitive analysis process dovetails into our campaign targeting work.
Display campaign audiences are the primary means for targeting display campaigns.
There are scores strategies for selecting display campaign audiences to fit the market targeted. The general idea is to understand the demographics, persona, and behavior of the customer; then to find or create audiences that match those characteristics. In addition to positive targeting, it may also be beneficial to exclude certain audiences. These "negative audiences" can be used to shape the targeting to improve campaign performance.
Types of targeting available include direct website placements by specifying URLs, so-called "in-market" audiences (behaviors indicating shorter-term buying interest), persistent interests known as "affinity" audiences, customer list audiences, and demographics audiences which might include age, gender, income level, homeownership, and parental status.
Because display ads are "advertisements" versus direct response marketing, the performance of display campaigns is even more dependent upon, and sensitive to the design of the ads, as compared to search ads.
The display ad channel is fundamentally different to search ad channels. Display ads deliver advertising messages, versus search ads which contain direct response marketing messages. We're showing display ads to audiences who likely may have a related need or interest, but haven't expressed search intent.
Search ads are display on search engine result pages, which the person searching has requested. Display ads are shown on websites where users are not necessarily expecting them. That's quite different from search ads where the user is searching for what you offer.
We've tested different creative approaches to invent, improve, test, and optimize display ads. Our process is based upon the classic creative process, mapped into digital advertising.
Display campaign optimizations take a bit more time to optimize. Many require 2 to 4 months, as different targeting methods are experimented with before finding a winning formula. Simpler display campaigns, and campaigns operating in business sectors with limited competition, can often be optimized faster. During the fine optimization period, it's not unusual to make significant breakthroughs in performance as we become more familiar with the nuances of how a display campaign serves, and the increasing volume of data makes it possible to test new ideas.
© 2023 Blastoff Labs™ | All Rights Reserved | Privacy Policy | Terms & Conditions
