“Half the money I spend on advertising is wasted; the trouble is I don't know which half.” - John Wanamaker
Advertising has experienced significant development since its inception, that said, now more than ever marketers want to better understand how budgets are being spent. The challenge is met by advertiser through attribution.
On December 5, 2018 iProspect collaborated with Facebook to showcase best-in-class case studies, where Facebook’s insight tools have been harnessed to measure incremental gains for our clients. The event explored three key pillars of measurement – lift, conversion and Attribution. This blog will focus on how iProspect have been working with a client to help it understand attribution.
What is Facebook Attribution?
Facebook Attribution can be accessed via the Measure & Report tab in your Business Manager Account and provides the following:
a people-based view of custom conversion events, using the Facebook Pixel,
cross-device conversion event mapping,
·various attribution models, including a data-driven model,
a 90-day look-back window; and
connectivity with non-Facebook advertising platforms like Google search.
Using Facebook Attribution with a key client
iProspect are currently working with a well-known shoe retailer that specialise in stylish and comfortable women’s shoes, born from a fusion of science and design. Using multi-channel digital marketing to drive e-commerce sales and grow their presence in emerging markets, their digital strategy is activated across several platforms within the Facebook family of apps, targeting audiences throughout the funnel.
The primary data source for the measurement of all digital channels has been Google Analytics 360, accompanied by the Facebook Pixel, and reported on a last-click attribution model. The Facebook Pixel is reported on a 28-day post click and 1-day post view attribution window, with Google Analytics operating to a lookback window of 30 days.
We knew that last-click attribution was likely to be overvaluing bottom funnel activity, but we didn’t know by how much. Furthermore, due to the view through restrictions within the Facebook family of apps, we knew that the full-funnel value of the platform was not being reflected in our Google Analytics 360 Attribution models.
What did we want to test?
We wanted to understand how the brand’s ads on Facebook, Instagram, Audience Network and Messenger are contributing to overall sales, both from a full-funnel and multi-channel perspective.
We started with two questions:
What is the true role of Facebook in the user’s journey and does that differ than suggested by incumbent measurement tools?
If we took a full funnel view of Facebook activity, how would this affect our channel investment?
How did we utilise Facebook Attribution?
Further to being invited to beta test the Attribution solution, our first step was to ensure connectivity to all necessary data sources; this included our Facebook pixel, purchase events, and 3rd party tracking sources for the mapping of data across other non-Facebook platforms. To enable us to accurately measure the incremental value Facebook’s activity, we selected the data-driven Attribution model within the tool.
Once we had a robust amount of data, we began our analysis.
What insight did we gain?
Facebook touchpoints are 3-4x higher in mid to upper funnel conversion journeys, with last-click undervaluing Facebook when representing it as a touchpoint just before a conversion.
Cross device engagements featured significantly in customer journeys, with 45% of conversions on desktop happening after customers interacted with Facebook ads on mobile.
The data-driven Facebook Attribution model shows that Facebook is being undervalued by 95% under the current last-click model in the platform.
A measured reduction in cost per acquisition (CPA) of 37%, compared to the data-driven valuations in GA360.
How did the insight change our strategy for the brand?
Initial analysis has shown us the importance of Facebook’s role in the mid-to-upper funnel stages of a consumer journey. As a result, we recommended increased investment to higher funnel campaigns within the Facebook platform, warranted by the subsequent reduction in cost per acquisition. Furthermore, whilst we continue to adopt a mobile-first approach, desktop remains front-of-mind when determining our creative strategy and optimisations; securing a presence across all-devices and a multitude of ad formats.
Finally, we will consider the feasibility of a multi-cell Conversion Lift test to better understand the causal value of these campaigns and the further validate the Facebook model.