Algorithmic Personalisation
The Background
Most personalisation technologies are rules-based which makes governing the digital experience difficult and limits the depth that we can personalise.
Australia Post wanted to explore the possibilities of automating personalisation. We created a test using advanced machine learning algorithms to create a personalised experience to answer the question- is automation worth it?
The Work
We launched an automated personalisation test within Adobe Marketing Cloud using an algorithm (random forest) to determine the best content to show a visitor based on their behavior:
- Test used no manual targeting rules – the algorithm personalised the content to each visitor as it learnt more about them
- To determine effectiveness of the personalisation, a 10% control group were excluded from the campaign, and only saw the ‘hardcoded’ homepage (regardless of what they did on the site)
- Test ran for a 2 week period
The Results
The test resulted in a 35% uplift in conversions for the variation
We’re now helping Australia Post make use of Adobe Audience Manager to use audience segments to scale-up automated personalisation efforts.
