As I answer these questions, the first obvious orchestration asset is the customer profile. In the broadcast days, marketers treated segments of customers, established marketing programs to reach each segment, and provided channels to support these segments without an explicit need to record the status of the campaign with each customer. Now, as we turn from broadcast to collaboration, the first question is, whom to collaborate with? A customer profile provides us with a list of prospects. As we proceed with collaboration, we must keep track of each of these customers and track their buying process by keeping track of their needs, their response to marketing campaigns, and their current understanding of our offerings.
In any enterprise, there are likely to be many views of prospects and customers. Most of the fragmentation comes from divergent views across organizations. While customer care organizations focus more on customer interactions, billing organizations track billing information. If my son and I are sharing a customer account, customer care may have knowledge of places he visits, as in service addresses, while a billing organization may only care about his billing address, where they send the bill. Customer master data management (MDM) solutions are popular ways of bridging views and bringing together a single unified view. However, over the past decades, this integration has been focused primarily on intraorganization sources of traditional “structured” data.
Most often in a structured data environment, the data must be integrated across a couple of organized sources, each carrying a customer ID as well as a customer hierarchy. As different sources are combined, they represent their IDs at different levels in the hierarchy and carry other information about the customers that can be used for merging customer data. For example, a customer care system that collects web clickstream data may deal with an individual customer and carry his/ her name, user identification, location of interaction, and so forth. The data in the billing system, however, may carry an account identification associated with a billing account for a household and the household billing address. A unified customer profile may carry many customer IDs from a customer care system mapped to a single billing ID from the billing system and, depending on the analytics requirement, the appropriate data can be organized and summarized from either of the two sources.
Usage data, which is described in chapter 3, provides the dynamic extension to the classic MDM-type solution. Unlike structured data, usage data may be used to create a number of customer attributes, which may change over time. For a quick-service restaurant, such as Starbucks or Panera Bread, loyalty card data can be used to identify a large number of customer profile attributes. These attributes may include usage preferences, locations, response to promotions, local weather patterns, and so on. In the location analytics example, I discussed mobility patterns, such as “work at home.” While the raw location data associated with a subscriber is fairly structured, these attributes and related microsegments are relatively more dynamic. Most marketers offering loyalty cards have begun to mine usage data to align additional customer attributes to static customer information available from customer service and billing. By adding usage data, we can start to differentiate these customers based on product usage, service location, time of day, day of week, or other significant attributes of interest to a marketer.
With the wide availability of social data, we have opened up the customer profile to also take into account social sources, including
Twitter, Facebook, Yelp, YouTube, other blogs, and in general any information that is publicly available. The information published externally could include intent to buy, product preferences, complaints, endorsements, usage, and other useful segmentation data. This data can be collected, collated, and identified with individual customers or segments, and connected with the rest of the customer view. How do we merge internal and external views to create what we may call a big data view of the customer? This integrated view is a far more holistic understanding of the customer. By analyzing and integrating this data with the rest of the customer master, we can now do a far more extensive household analysis. This data may reveal additional information about customer satisfaction with the product. For example, there may be low usage for a product because of lack of access, as opposed to disinterest in the product. The social media chatter may be able to discern geographical locations where scarcity is leading to lack of product usage. While this data can also be collected via better data collection at the point of sale or through consumer surveys, social media data collection may provide a more comprehensive sentiment at a lower cost, and could be far more dynamic in revealing spikes in sentiments and associated causes.
Another good source of data is customer profiles from other industries. As a retailer, I may have a good customer profile about my customers’ usage pattern. A telecom provider may provide complementary understanding of their mobility patterns, and a cable operator may have a good understanding of their media viewing habits. Mobility patterns are key to providing context-specific promotions to customers. For example, while the parents may be paying for the cell phone, the actual user may reside in a college dormitory in a different city and should not be offered promotions for regional stores in the city where the parents live unless the student is visiting the parents for Fall break. Figure 5.1 shows sample elements of a big data customer profile. It includes demographics, social patterns, buying patterns, and mobility patterns. How would a marketer find these jigsaw pieces and pull them together to get a holistic view of the customer?
Figure 5.1 Big Data Customer Profile
A number of cloud-based marketing organizations are keeping track of web usage patterns. By analyzing the websites browsed by a specific customer, these organizations are establishing their customer profiles and creating attributes like “interested in golf,” “stock investor" and “classical music lover.” Data management platforms combine this data with the past history of advertisements placed, viewed, and clicked, to generate a sophisticated understanding of a customer and his/her interest in a specific campaign, and the saturation level.
The customer profile must stay focused on its purpose. For example, the Obama campaign created a voter profile with two objectives— one to predict the likelihood of someone voting for Obama, and the other to predict the likelihood of someone contributing to the Obama campaign. This campaign began in 2012, the second-term election year, and tracked the name of every one of the 69,456,897 Americans whose votes had put him in the White House in 2008.3 They may have cast those votes by secret ballot, but the analysts could look at the Democrats’ vote totals in each precinct and identify the people most likely to have backed him. They started with a customer profile for
180 million potential voters and updated their information on a weekly basis to drive a series of electoral campaigns.