Let's talk about similar audiences
If there’s one thing we’re obsessed with, it’s audiences. Whether this is remarketing or targeting completely new users, we are constantly looking at our audiences to provide more insight into our data. Originally for the display network only, when similar audiences for search launched in beta earlier this year we were all super excited at Merkle | Periscopix to start using these in our campaigns. Now this has been officially launched for both search and shopping campaigns and is available to everyone.
Essentially, similar audience lists target your ‘favourite strangers’ by looking for people who behave like users in your remarketing lists (which could be, for example, people who are similar to your most valuable customers). Powered by Google’s machine learning, these new users exhibit a range of online signals which indicate to Google that they are interested in your product or service and therefore are more likely to show a positive engagement.
The above image is a visual representation of audience profiling. In the first icon we use criteria to pick out certain people and put them in a remarketing list (e.g. website visitors from the last 180 days) and these are the users who are in the second icon. The final icon shows people that are similar to those in the remarketing list but have never visited our site.
Sounds great! How do I implement?
Implementation is pretty straightforward as Google identifies remarketing lists that are eligible and will automatically create a similar audiences list. All you need to do is overlay the similar audience lists on to all of your campaigns, whack a 0% bid adjustment on and watch the data come through.
How can I optimise?
The lists will start to populate over time but there are a number of things you can do to ensure maximum potential:
Remarketing lists clear up: Similar audience lists are based on remarketing lists so make sure you have as much coverage in your account as possible by having a range of different yet useful lists e.g. past visitors for different duration times, past converters, non-converters.
Monitor: As the lists will take time to populate, keep a close eye on the data coming through. As this is a list of new users, key metrics such as CPA can initially seem high compared to the metrics for a standard remarketing list. It is important to regularly check the data to make the necessary bid adjustments.
Bid adjustments: As mentioned above, bid adjustments are key. It is best practice to start with a 0% bid adjustment so that the data is unfiltered. After a month or so, once you have enough data, start making bid adjustments based on the performance of these lists. For example, if conversion rate is higher than usual, increase your bids to appear more prominently on the SERP for these users.
Analysis: Go chart crazy! Analyse audience data vs non audience data to truly see what impact audiences have on your campaigns. This will help determine who you want to target and how your current targeting options have been performing.
What does good targeting look like?
At Merkle | Periscopix, we’re constantly analysing our audiences to ensure we are targeting the most relevant people. We recently analysed audience data over a 3 month period for a client in the estate agent industry. The table below shows a significant difference in click-through rates and how CPA is lower for users in both the Remarketing and Similar User audience lists. These users have shown a better engagement rate and have a higher likelihood of converting on the site. This can then be fed back into campaign strategy and used to inform bidding and targeting strategies across the account.
In summary, the best thing to do is overlay similar audiences across all of your campaigns with a 0% bid adjustment, wait for the data to come through and after a few weeks or months (depending on traffic volume) - make an informed decision on whether to bid up, down, remarket or exclude specific audience lists.
More information can be found on Google’s official blog.