Audience data is key to a good PPC (pay-per-click) strategy, and it can be collected from multiple channels. Someone watching or liking one of your videos on YouTube might be used to inform bidding strategy in search; the content you show someone on a display ad could be informed by the landing pages they’ve visited, or by the meta tags of the video they’ve clicked on.
There is much to be gained from a multichannel strategy. The difficulty is getting that strategy right. Here’s how we go about it at our company.
Multichannel retargeting
Working with a client in the education sector, we have developed a way of integrating search, display and video, primarily in terms of retargeting.
First off, it’s likely you’ll need to do a thorough review of the audiences in your account. With remarketing lists for search ads (RLSA), Google Analytics, Customer Match, mobile app and YouTube audiences available, it’s likely there are many historic audiences which are just over-complicating the situation and destroying the transparency you had visualized by too much segmentation.
Once you’ve refined your audience strategy to include the high-intent audiences (all converters, abandoned basket and so on), product/category audiences, page engagement (>2 mins on site), all visitors and, of course, segmented your Customer Match data in a way that will be most beneficial to your strategy, it’s time to work on mapping out a matrix.
Search –> Search
Layering your audiences over the top of your search campaigns should be the first step for all advertisers. Be careful, though, as it’s easy to get carried away here and just layer on every audience you’ve ever made, and their corresponding “similar to” audiences as well. However, it’s better to be clever about this: Many of your audiences won’t get enough volume to be useful, and it’s likely that they won’t all be used anyway due to Google’s algorithm prioritizing the audiences by the bid modifiers applied.
In general, if you have a very generic campaign, it’s best to layer more specific audiences, as you’re less informed about their intent. And consequently, if you have a product campaign, you can afford to layer over more generic audiences, as you already have more insight into what these customers are looking for.
Once you’ve got the audiences in the account, don’t just leave them there! Analyze the data, apply bid modifiers, and make sure to keep monitoring performance and making changes so that you always have the best modifiers applied. Also, by using IF statement ad copy, you have a really easy way to modify your ad copy based on the audience. Ideal!
Search –> Display/ YouTube
Data from search activity also can be used to inform retargeting on display by way of bid modifiers and ad content. Audiences that appear to have high intent, for example, might be shown ads designed to drive purchase; those with lower intent might be shown content aimed at building awareness. Intent should also inform the size of the bid, with higher intent usually leading to larger spend.
Certain website interactions might also shape the content of display ads. If someone has shown interest in a particular educational course, they can be shown an ad that specifically relates to it.
The same can be applied to YouTube, though there tends to be less flexibility in terms of varying the content of the ads — simply because it takes longer and costs more to make the videos.
YouTube –> Display/Search
It’s very similar when you retarget in the opposite direction, except you can’t really assess level of intent via YouTube. Instead, audiences can be created according to the videos or ads they’ve seen, channel subscriptions and interactions with the videos (likes and shares).
The display ads can once again be adjusted based on the apparent interests of a particular audience, and the implied level of intent. For example, a subscriber to our client’s channel, combined with a high level of interaction with their videos, may be shown lower-funnel ad content as a result.
Within search, bid modifiers reflect level of interest in the brand. Someone who has already seen a lot of YouTube ads is more likely to convert, as is someone who has liked several videos. Ad copy can also be adjusted, depending on the subject matter of the video they saw: the name of a specific course, a reference to a case study mentioned in the video, or a particular aspect or theme of their business.
Layering audiences
So far, it may all sound quite straightforward. The tricky part is creating a method for selecting and combining audiences.
In this example, we were able to identify more than 400 individual audiences in our brainstorm, and therefore millions of possible combinations of those audiences. To cut these down, you can first apply the MECE (Mutually Exclusive, Collectively Exhaustive) approach.
Some audiences overlap — for example, “Visited Site 24 hours ago” and “Visited Site 3 days ago” — so your custom audience will need to include an exclusion. There are also datasets which might be cut altogether, as they are either not very significant in terms of measuring intent or because a similar audience basically covers it.
It helps to visualize users according to their place in the search funnel and where audiences fit within it. Once you have this in place, it is easier to create a hierarchy of audiences and therefore to know which ones to exclude. This can be partly data-driven, but it also relies a fair bit on common sense. A blog subscriber is likely to be lower in the funnel than someone who has recently made it to the pricing page.
The benefits
Adopting a multichannel approach enables more effective bidding, greater personalization, and it can be used to inform your overall marketing strategy. AdWords custom audiences can include user behavior from search, display and video to add a level of granularity previously inaccessible within the platform.
Though we are not able to publish specific performance data from our client, the uplift in conversion rate and improvement to CPC (cost-per-click) has been considerable. And there will be plenty of opportunities to develop this strategy even further, especially as Google is constantly rolling out new betas in this space.
Opinions expressed in this article are those of the guest author and not necessarily Marketing Land. Staff authors are listed here.
About The Author
Passionate about both advertising and data, Grace Kaye has become a specialist in programmatic online display where she uses programmatic as a vehicle to provide excellent user experience whilst satisfying business goals. Grace now leads a team at Brainlabs delivering high-performance display campaigns where precise audience targeting, data-driven optimisation and high-quality content is at the heart of every strategy.