Data Issues Require Perspective

Share this article:
It is sometimes difficult to describe the massive volumes of customer data that the average business-to-consumer heavyweight company now generates and stores.


The customer relationship management market is maturing. For one thing, it has a fairly consistent name these days, and the role of gathering and analyzing customer data to improve customer relationships is relatively well accepted. However, now that the majority of BTC organizations routinely make some attempt to make sense of their customer data, companies that want to maintain their competitive edge have to find ways to get this understanding better and faster.


Recent discussions regarding CRM have focused on personalization and the operational systems and infrastructures that allow BTCs to communicate with vast numbers of clients far more efficiently across a range of channels. With a developing ability to implement outbound campaigns in ever-shorter time frames, companies must focus on making those communications more effective - otherwise they run the danger of simply doing the wrong thing in real time. This is why analytics for CRM are emerging as an accepted part of the CRM mix. The problem is that many tools are still relatively basic compared with the technology around them.


While Web servers, phone switches, electronic point-of-sale and ATM systems collect millions of pieces of data daily, too many companies that invest millions in these operational systems struggle to turn such raw transaction data into new information on the behavior of each individual customer.


Most practitioners are restricted to sampled or aggregated data sets that can be hard to construct and often do not represent all the hues of the customer base.


This problem becomes far worse when marketers then try to select subgroups for differential treatments. Keeping these segments large enough to draw sensible conclusions is vital. There are two technology features that can bring more dimensionality to customer analytics. The first is visualization, which has long been talked about, but has yet to permeate the business world as much as it should. The second is computing scalability, which is rarely addressed and often dismissed as something overly complex that should stay in the realms of the technical savants.


However, the reality is that the software and hardware do exist - today - that allow marketing analysts direct access to entire multimillion-person customer bases quickly, easily and affordably. By retaining the granular detail of the customer database, analysts have the power to interrogate the data from any angle, zoom in to the individual, take a step back to view groupings and zoom out to see the whole picture. Every vantage point has something different to offer and brings a depth of perspective and better understanding of what is really contributing to the bottom line.


Scalability and visualization are natural partners. The ability to analyze large volumes of data is somewhat diminished if those results cannot be presented in a simplified yet insightful way. Interactive visualization allows multiple dimensions of information to be condensed into one screen. Scalability adds the ability to click through immediately to areas of interest or to pan back to see a complete customer base in one view.


Scalability brings an added degree of accuracy to the customer behavior modeling process, with a better view of extremes of behavior (the most profitable, the least risky) and the ability to create more robust segmentations that are large enough for statistically significant returns. The insight that visualization offers is compounded by the speed at which that insight is delivered - speed that supports the operation of business in Internet time. The ability to take a look at 15 million, 20 million or even more customers in one glance helps to make sense of the considerable investments in operational infrastructures.


Zooming out on your customers can be a great sanity check on all sorts of data issues. For example, consider the following true story about a bank and what happened when it took its first look at all of its customer data.


The bank wanted to extend credit card ownership among its customers and - after some basic data exploration that gave it some insight into the general distributions - the bank sensibly decided to take stock of current credit card ownership patterns within its customer base. Plotting the entire customer base on a geographical map, it was immediately apparent that the bank's acceptance of credit applications was heavily biased toward two geographical regions. A moment of awkward realization followed. The regions corresponded to the two main locations of the bank's offices. Acceptance criteria were being governed because data on its own staff (naturally low-risk) were biasing the statistical models to classify corresponding geographies as inherently good.


The credit scoring system had created a highly statistically accurate model that identified the bank's staff members, with their near perfect credit ratings, as the best risks. But no one had lifted their nose from the grindstone to take a wider perspective. It was a very costly mistake in terms of missing ideal targets for the product - just because they lived in the "wrong" town.


Analytical tools have come a long way during the past few years, and the rash of acquisitions of analytics companies during the past 18 months shows that analytics are fast becoming a "must have" rather than a "nice to have." While business intelligence, statistics and desktop tools all do an important job and will continue to, the market needs to look one step beyond.


New marketing techniques and processes need new marketing technology. Customers are a living, ever-changing resource, unique to each organization. No BTC should be content with relying on a canned view of customers. The competitive edge will go to those that leverage their investments in operational infrastructures and demand tools that can match the volume and multiple dimensions of their customer base.


• Mark Smith is president of Quadstone, Boston, a customer behavior modeling and predictive marketing software vendor. His e-mail address is mark.smith@quadstone.com.
Share this article:
You must be a registered member of Direct Marketing News to post a comment.
close

Next Article in Data/Analytics

Sign up to our newsletters

Follow us on Twitter @dmnews

Latest Jobs:

Featured Listings

DataMentors

DataMentors

DataMentors is a full-service provider of data-as-a-service, data management, business intelligence, and campaign ...