With the increasing publicity of digital economy, the need for applications that pushes this trend is also important. First let’s list down few of the digital trends that could trigger the digital economy in future; globally people use Twitter, Facebook, Pinterest, LinkedIn, etc. These are just a few of the highly popular data sources. Data emanating from many of these applications are enormous and user specific. The data sources mentioned above also exposes channels to collect this data at ease. Each of these sources has a specific data type and behavior. Micro blogging sites gives out sentiment data, Facebook gives out opinions, Pinterest give out product ideas in pictorial format, LinkedIn gives professional status quo of a user. The list goes on. Now, can I make these part of my digital economy?
Now let’s abruptly switch to another application domain; CRM (with focus on sales function). Conventional CRM application contains data and functions about customer master data, list of products and promotions that a company intent to showcase or sell and the sales opportunity of these products with customers who are its master data repository. Isn’t this limiting the growth prospect of the company? How can a company increase its customer base? How can a company improve the product innovation? How can a company keep pace with changing customer behavior? Even in the so called digital economy, company spent considerable amount of money in advertisement, road shows, door to door sales etc. Can they reduce these spending and put the money in other productive venues? The answer to these questions is “Yes”. Customer Relationship Management (CRM) has to change to another paradigm called “Digital Economy Marketplace” (DEM). Each and every product that is produced in the factory and listed in the DEM, has to be viewed as “Made to Order” and “Sold to Intended”. A product must not wait in the shelf of a physical store waiting for an unknown user; rather a user must be informed about the availability of his desired product, the moment he tweets about a convenience feature he is expecting from a product, or review about a competitor product that he has just purchased. We all are aware about the custom built Ferrari and Lamborghini. Each and every piece of these machines are custom built to an order. The same business model is applicable for medium to semi-medium end consumer products also.
In the old age economy millions of products, worth trillions of dollars are waiting in store shelves, expecting a customer to visit, pick and buy. Nobody is sure if those will be sold at all in a given time window. Many of those products cannot be refurbished. This leads to lose for capital to manufacturer and all others in the supply chain. Many product innovation gets a buy in from customers, only after periodic trials in market. The amount of sample products (if durable) that need to be manufactured and distributed for trial cannot be precisely predicted. Amount of recurring customization that customers expect in standard product cannot be predicted. All these involves lot of investment for trials. Some of the trials could eventually translate to huge success and some may fail.
The first step in reducing these loses is via finding the right user and user interest. How can a company find the depth of acceptability of its product? How can a manufacturing company find a right user if the customer data that it hold in its digital repository is purely informatic in nature? The textual nature of a customer data (his name, address, salary, past purchase information etc.) need to have a dynamic behavior. Dynamic nature can be added to a customer data via a behavioral contextual info. As the end user physically grows, his needs changes, his behavior changes, his buying capacity changes, his social status changes. Most of these could be a dynamic contextual data that changes every day. This info cannot be collected from physical store or door to door enquiry. But this could be derived from the online tools and applications that we discussed in the beginning of this blog. Second step of reducing storage cost is by refurbishing digital data (of products) rather than physical product. Display and marketing of trending product must first happen in digital format, rather than physical format. Third step is by virtually collecting and adapting innovative ideas and feedbacks from technologists who constantly tweets about technology that make up these products.
SAP is ready for this digital commerce revolution using a complete set of tools, technology and applications. The following architecture describes the architecture of a mash up application that supports digital market place. SAP Hybris is core application and architectural pillar in this mashed up architecture. Hybris’ core architecture provides a master data-management layer to ensure consistent inventory, pricing, order-status and other information across channels, whether that’s Web, mobile, call center or retail stores. There’s also a process-management layer that applies the same business rules across channels, so prices and promotions encountered online are consistent with those encountered in stores or on mobile devices. Continuous modification of dynamic behavior of customer master data (which is needed for identifying the right users for a company’s product) can be achieved by accessing the tweets and other data from all the three data sources mentioned (Twitter, Facebook, LinkedIn). Reference Figure 1.
Custom built components that listens to the tweets accesses customer sentiments, comments and his social status using the public API exposed from all the three data sources. This data is dumped into a HDFS data storage. SAP HANA Vora is used to query the data that is available in the HDFS data store (in real time). The in memory architecture of Vora enables fast and real time query between big data storage and HANA customer master data for accurate data collection. These data could be co-related using the personal mail id that is commonly shared between these two systems. Once the data is co-related, SAP Predictive analytics is run to determine if there is a chance for a customer to buy a specific product that is either in digital store or that is in innovation center of the company. If the prediction confidence score is more than 90%, then the same product feature is displayed in the customer’s portal or mailed or send as a message to the mobile that is registered with the portal master data repository. Let’s now review few use cases that has under gone testing of this architecture.
Case 1: Alex is a frequent flyer and visits many countries for business purposes. Due to his busy schedules, he seldom visit brick and mortar stores to buy his travel necessity. Last time he had brought a travel bag from commerce portal abc.com (his frequent buyer portal). But he is not happy with the quality of the lock, due to its frequently jamming click and slide system. Last day he got annoyed of this and tweeted about this. Our architecture got hold of this tweet, analyzed and recommended a better product; next time he logged into “abc.com”. Alex was surprised and excited with the proactive approach and subsequently the loyalty towards “abc” increased.
Case 2: Max is an ardent car lover. He changes his car every 3-4 years (every time he get promotion) and tries to add lot of interior features. He is not sure about what are the latest technological advancements. Company P wants to give a preview of its new car to Max, before it launched. Company P, who is using the above application came to know about Max’s social status from the LinkedIn update he had done few days back. That effort eventually translated into a confirmed pre-order.
The applicability of the architecture and features are innumerable in the digital economy.