I’m trying to do an attribution model for our marketing. We have markov paths for our facebook conversions, but owing to the post-Cambridge Analytica facebook data sharing issues, we don’t have data on non-converting paths. Further, our data compiler is providing the data at an aggregated level like this:
|01/08/18-07/08/18||Facebook-[Brand Campaign]-Click> Google-[Brand Campaign]-Click||50|
|01/08/18-07/08/18||Facebook-[Brand Campaign]-View> Google-[Brand Campaign]-Click||20|
|01/08/18-07/08/18||Direct-Click > Facebook-[Campaign Name]-Click||1|
|01/08/18-07/08/18||Direct-Click > Email-Click > Direct-Click > Facebook-[Campaign Name]-Click > Email-Click > Direct-Click||1|
My understanding is that the R Channel Attribution model for Markov channel attribution requires non-conversion paths as well as converting paths. Has anyone found a way around this?
I have suggested doing top-down marketing mix modelling but the marketing teams are particularly interested in when in our customer journey facebook is most useful.