Introduction:
Over a year ago, Google offered a new way to export GA data into BigQuery, that is streaming data in a near real-time fashion instead of the batch updates 3 times a day. We thought of many scenarios where this can be helpful. We learned a lot through these discussions and implementations throughout last year.
Analysis Overview:
Here are the most common use cases:
- A retailer was interested in creating a flexible alert system that sends out e-mails and/or SMSs when the number of sessions, pageviews, or transactions goes below a certain threshold compared to the same day and hour of last week.
- Another retailer was interested in building a dashboard showing key measurements, such as number of sessions, pageviews, and transactions, by the hour on a live dashboard, to monitor the site performance.
- Another client, a video site, was interested mainly to get the user behavior data in a near real-time fashion exported, to process it further, and apply predictive analytics to it. The client could then do several things with it, such as:
- Detect users who are about to churn, unsubscribe, or leave the site and offer them a special discount, i.e.instead of giving away the discount to users who are staying anyway.
- Personalize the user's experience based on his previous watch history or based on clustering look-alike users and recommending to them videos that they are likely to be interested in.
All of these scenarios have a great impact on the user experience and ultimately on the business.
We'll focus on the second scenario above: creating an hour-by-hour live dashboard. We will also try to touch on the other use cases whenever possible.
Dashboard showing number of session and pageviews from real-time feed compared to previous week.
DAA members, go here to view full recipe.
------------------------------
Hazem Mahsoub Soliman
Data Engineering Manager
E-Nor (Corporate Account)
Santa Clara CA
------------------------------