Our attribution model is custom built by our clever data science team. We use a multi-touch, incremental approach, using a neural network which takes 200+ inputs which are unique to each client. Within the platform, we also display last-click and first-click performance, but mostly for comparison performance against our multi-touch approach - we find this especially useful during onboarding.

Using sync IDs, Cubed stitches together data from multiple devices, providing a holistic and accurate view of the customer journey – this is key to understanding the performance of all marketing activity.

We re-train our attribution models weekly, so your models will always by up-to date with the most recent data.

We have the capability to upload offline sales into our system. We use our syncIDs to stitch together offline sales and online activity where possible.

Cubed attributes conversions and revenue at the visit and page level (the most granular) so we can truly understand the impact of each marketing touchpoint. We build our models using neural networks due to their capability for complex pattern recognition and adaptability to varied data types, offering a more nuanced understanding of customer journeys.

This granularity of attributed data unlocks the potential for things like campaign and content attribution analysis, alongside the out-of-box channel attribution due to the ability to aggregate the data up. Markov or Shapley struggles to attribute without having a huge dataset for training, in addition to being computationally expensive, resulting in attribution for smaller business not producing reliable insights.

While Google Analytics and Adobe models like Markov or Shapley have their merits, neural networks offer a more advanced and versatile solution for businesses seeking a sophisticated and adaptable attribution analysis approach.