Attract, Observe, Recommend, Analyze: Supporting Retail Campaigns with Data Analytics
By Henry Lee, Solutions Architect, and Charles Tse, Staff Solutions Architect
Campaign data analysis is important not only to understand event effectiveness, but also to support offline store intelligence such as shop/booth layout design, dropout ratio in relation to customer flow. It provides another channel to observe customer demographic and buying behavior. With all these information, we can provide product recommendations with a higher level of confidence.
To drive the event with data and AI, we can follow 4 simple steps:
Attract, Observe, Recommend, Analyze
Customer Attraction
For customer attraction, we see a number of retailtainment technology being used such as AR that is applied to “magic mirrors”, or interactive games or company apps. Smart shelf/vending machines for promotional products in popup shops, which may also be integrated with product recommendation to provide personal advice.
User Behavior Observation
All these attraction channels include online/offline interaction with customers. It provides a good chance for businesses to observe customer behavior or to better understand their interests. The observation can be related to online comments for the event activities, coupon redeem ratio, customer flow track in popup shops or demographic recognition both online and offline.
Through attraction interaction points and observe process, correlating history customer behavior information. We are more confident to provide product recommendation in the campaign.
Product Recommendation
We can generate product recommendations based on 3 types of data inputs with Alibaba Cloud’s Machine Learning Platform for AI (PAI) service. The data types include customer demographic, product information, and customer behavior such as purchase history. With customer demographic, it gives us another way to provide recommendations, independent of history, which can be beneficial when dealing with new customers who have not completed a transaction before. We can provide single item recommendations, top items recommendations, or item-to-item relationship suggestions for different scenarios. Recommendation results can be used in company websites, mobile applications, or booth devices in the event.
Recommendation models can also be fine-tuned with customer transactions in the event, improving the hit rate of the data model and supporting us to provide a better model for future events.
Data Analytics
Popup store would be another good observation point for data analysis. One of the most common measure would be measuring user behavior through camera devices. This supports not only the visitor count, but also help to capture basic demographic information such as age range and gender. Another part of data analysis would be on route and counter stay time within popup shops. We can also perform more detailed analysis such as estimating the conversion rate of purchases by measuring the number of visitors to the cashier area. All these can help us to understand the customer flow and measure the effectiveness of attraction points.
To prepare for future campaigns, we can also reference on the event data to locate high value customers. We can target those who are responsive or active in an event and those with high conversion rates. With advanced data analytics and big data, we can understand the users’ common demographic and look for similar (lookalike) customers.
Conclusion
This article only lists out part of the possible areas for data analysis in retail campaigns. With better technology and innovative ideas, we could do even more to drive events with data. All these ideas are only possible with a powerful, cloud-based platform to support the data analysis. Learn how Alibaba Cloud helps you achieve this with our Machine Learning Platform for AI (PAI), MaxCompute, and QuickBI platform to have a quick start on your campaign data analysis.
