Various pieces of research (see here, here and here) have argued that social media’s role in the purchase path is undervalued by the last-click attribution model, which remain the preferred attribution model among marketers. A new study [pdf] from Kenshoo Social indicates that Facebook ads similarly suffer from undervaluation in last-click models when compared to a variety of multi-touch models, by up to 30%.
The researchers measured millions of converting click-paths across a wide breadth of clients via Last Ad measurement, running those same conversions through several standard multi-touch attribution (MTA) models. Relative value was determined by looking at cost-per-acquisition (total media cost divided by total conversions).
The results, sorted by attribution model, are listed below.
- “First-Only” attribution model – whereby only the first touch is credited. In this case Last Ad models undervalue Facebook by 30% relative to this model.
- “Prefer First” model – where the first ad gets the majority of the credit, and the rest receive credit in a linearly decaying fashion. Facebook ads are undervalued by 20% by the Last Ad model relative to this one.
- “Divide Equally” model – where each interaction gets equal credit for the conversion. Compared to this model, Facebook ads are undervalued by 16% by the Last Ad model.
- “Prefer Last” model – where the last ad gets the majority of the credit, and the previous ones see credit in a linear-based decaying fashion. Last Ad models undervalue Facebook by 12% compared to this model.
- “U-Shaped” model – where the first and last clicks together get 80% of the credit (40%) each, and the other 20% is split evenly between the middle touches. Compared to this model, the Last Ad model undervalues Facebook by 15%.
The finding that Facebook is undervalued more in the Last Ad model relative to models that emphasize the first interactions suggest that Facebook ads see more interaction earlier on in the purchase path. Such a reading would be consistent with recent data from Google, which found social media to play more of an “assist” role in the online purchase journey.
About the Data:Â The research reflects an analysis of live campaign performance data for Kenshoo clients managing ads across multiple digital marketing channels during March, April, and May, 2013.
The large brands included in the study span a wide range of industries including retail, home improvement, and financial services. The total data set covers hundreds of millions of clicks and millions of direct online sales conversions Even though MTA models are able to recognize the value of ad impressions and their positive impact in driving conversions, for the purposes of this study, only clicks were analyzed across customer journeys.