Here are some projects I deem quite cool that I have built in the past
Realtime Analytics and Forecasting Platform
- Problem: With a very unstable website, we wanted to use the real time capabilities of Adobe Analytics to understand technical performance and sales impact. It needed to be a flexible platform allowing for future extension and evolving forms of data representation.
- We chose Elasticsearch as the storage backend. It is fast, scaleable and resilient with relatively low maintenance
- Using different scripts, we pulled real time and batch data out of Adobe Analytics and stored in ES
- To get monitoring data, we rolled out Dynatrace and pulled the data into our platform as well
- Once the data was available, Python was used to build a precise forecast on historical data, which was then also put into ES
- Grafana was then used to visualize data on desktop, mobile and TV screens
- The same data was also displayed on LaMetric time smart displays and made available to an experimental Alexa skill
Digital Marketing Attribution Modeling
- Problem: To better understand our Digital Marketing performance, we needed to move past standard last-click attribution. In addition to web analytics data, we incorporated other digital marketing data to complete the customer journey with offsite touchpoints.
- To get data out of Adobe Analytics, a raw data feed was set up to a FTP destination
- NiFi was then used to parse the data and store it in Elasticsearch
- We setup scripts to iterate through that data and rebuild the customer journey with custom defined touchpoints and visitor logic
- The results were stored in Elasticsearch and visualized with Grafana
- Using classification, parts of the result were imported to Adobe Analytics for easy access