The past few years have seen historic increases in online sales. In fact, eCommerce sales rose 32.4% in 2020 alone, the largest one-year increase ever recorded for online sales, and more than 57% of consumers now prefer to shop online. At the same time, 56.8% of the global population is now engaged on various social media platforms. Considering both trends, retailers are beginning to launch new retail channels on social media platforms to enable a seamless and convenient shopping experience without redirecting consumers to an eCommerce site. But building a social marketplace requires a new business model, as well as changes to traditional processes of inventory management, payments and logistics.
Social Selling has Specific Requirements
Social selling has unique requirements. The focus is on item visibility and promotion, real-time inventory updates and reports, and integration with payment engines and logistics providers.
Figure 1: Social selling workflow
Enhancing item visibility and promotion is an important aspect of the social selling process, and it requires the use of AI/ML algorithms. These technologies position an item in a user’s feed based on context, location, preferences and other criteria. They can also track customers’ behavior across social media and eCommerce portals to determine preferences and feed them content that matches their likes and needs.
But for social selling to work properly, changes to inventory data from databases will need to be captured and provided to the marketplace in real-time, along with relevant user profile information. There are a few approaches to capturing changes to data from databases. These approaches include adding columns with the last updated timestamp, defining database triggers at the row level to fire on each insert, making updates, tracking deletions and performing log-based change data capture. Each of these strategies comes with challenges or limitations, from increasing database overhead and decreasing performance to understanding the database logging formats and metadata changes.
A Solution for Real-Time Selling
Figure 2: Real-time social selling process flow
Change data capture (CDC) is an approach to overcome these challenges. Instead of other intrusive methods, the CDC relies on polling tables, triggers and transaction logs. One of these, transaction logs, can use the existing transaction log files that already exist for database backup and recovery purposes. The transaction log files contain the insert, update and delete changes across all database tables, and can be used as a source for these changes. The downside of this approach is that different databases have different standards and formats.
The Debezium open-source platform can overcome these challenges by providing connectors to a variety of databases and the ability to read different log formats. These connectors can be configured to watch the inventory database for changes and publish them into Apache Kafka topics, which can then be used as real-time information for item positioning.
Next, the most relevant content needs to be sent to social media users. Kamelets are easy-to-install Apache Camel K connectors that offer plug-and-play reusable connectors for external systems without having to writing code. These connectors can be used to read from the Kafka topics and send the inventory information to the social media marketplace. They can also provide critical integration to send customer profile data from a CRM application and behavioral data from marketing applications to AI/ML applications. They also integrate with logistics providers for shipments. This solution provides the customized feed that customers expect, and an overall better customer experience.
Social Media Emerges as the Next Marketplace
Social media marketplaces are becoming new retail channels that will provide customers with a safe and convenient way to shop online. But the implementation of these requires changes to traditional retail business processes and applications. Companies can use new technologies such as Debezium, Kafka and Kamelets to enable a real-time integration backbone that makes it easier to implement these new social selling retail channels.
