With different attribution methods used to measure different channels’ performance, it can be difficult to measure the true impact of our advertising. Google Analytics (GA), allows us to track multiple channels and pool information together to measure performance. It is one of the most common tools used by advertisers to achieve this. This means it is vital that our Facebook campaigns are set up to generate strong results within GA.
While GA supports several attribution models, we will be looking at ‘Last Non-Direct Click’ (where purchases are attributed to the last channel to receive a click before the user converts on-site) and what you can introduce to your Facebook planning and activation to drive strong paid social performance, while helping improve cross-channel performance.
Matching metrics and events between the platforms
When our data pulls into GA, traffic to site is measured through a ‘Session’, which tracks a group of interactions a user takes within a given time frame on site, contrasting with Facebook’s ‘Link Click’ metric, which registers when a user clicks on an ad’s link. Therefore, exploring other optimisations can prove beneficial.
Within the ‘Traffic’ objective on Facebook, we found that the ‘Landing Page View’ (LPV) optimisation led to a 50% decrease in cost per sessions compared with link clicks.While optimising a conversion campaign to ‘View Content’ drove slightly less cost-effective sessions than LPVs, it generated more conversions, making it more suitable if ROI is a secondary KPI to cost per session.
Purchases tracked by the Facebook pixel are typically attributed over a longer period (up to 28 days), compared to GA’s Last Non-Direct Click’ attribution.
Getting your targeting perfect
Make sure that your targeting drives the best performance on your platform and generates great results on GA. One example would be utilising Facebook ‘Lookalike’ audiences, which creates an audience of users with similar traits to users performing a specific on-site action (i.e. purchasing a product). You can go on to test the lookalike size (measured through %). Initially we felt a 1-2% lookalike audience would perform stronger due to ‘relevance’, however, after further testing, we found that 4-6% lookalikes yielded stronger results. The same can be applied to custom audiences, where you can expand the audience by increasing the lookback window.
Be mindful of audience size and relevance. The longer your campaigns stay in the ‘learning phase’ (where Facebook requires a certain volume of conversion events based off your optimisation), the less likely it is to drive cost-effective sessions. This is particularly higher risk for ‘lower-funnel’ targeting, where your audience pool may be more restrictive.
When confident in your audiences, consolidating those into one ad set can lead to stronger results. The conversion events required within the learning phase are broken out by ad set. Merging audiences to reduce the volume of ad sets will provide more data (and more budget) for those remaining, allowing you to move out of the learning phase faster.
