Customer Centric Data Refinement in the Age of AI and Machine Learning by Stephen H. Yu [Video]

ANA Conference 9/25/2024

Advanced analytics is within every marketers’ reach thank to rapid advancements in modeling techniques based on machine learning and AI. Nevertheless, even machines prefer clean data. Especially when holistic personalization is one of the main marketing goals, properly managed customer-centric data will increase the effectiveness of model development, deployment efforts, and backend analyses.

Development of a customer-centric database requires a comprehensive data roadmap, as a Customer Data Platform is not just a pool of available data in one place. All data elements must be carefully converted into “descriptors of individual” for each target. Creating a Customer ID system through commercially available services is just the first step.

Further, all data elements must become “analytics-ready” for effective segmentation, modeling, and targeting, even when the bulk of such tasks are done by the machines. It has been quite common that data scientists often spend over 80% of their valuable time in pre-analytics processing such as data cleansing, categorization, and summarization. Such work should be done “before” any serious analytics efforts begin, and resultant “Analytics Sandbox” will expedite all analytics efforts from pre-testing to post-deployment studies and attribution.

Key Takeaways:
– Why marketers need “analytics-ready” customer-centric data depositories in the age of AI
– What modern CDPs must be
– Types of pre-analytics data work for 1:1 targeting and personalization
– Design concept of “Analytics-Sandbox” for advanced analytics
– Creating effective data menu for CDP development
– Data summary guidelines for transaction data and other behavioral data
– Data categorization rules for seemingly messy data fields such as product, offer, channel, source, etc.
– Check-list for 3rd-party data procurement

Speaker:
Stephen H. Yu, President & Chief Consultant, Willow Data Strategy, LLC