Performing New Tricks for Old Customers

Before I arrived at Cisco Systems in 2001, the company was focused on building things for a rapidly growing business.

When the economy changed and that growth diminished for a period, the focused turned to: “How do I become closer to my customer? What do I need, from a data standpoint, to truly understand my customer’s status, situation and needs?” The answer: a single view of our customers.

Cisco has thousands of customers. Interactions with them, from requests for help and “open-problem” tickets, were stored in different databases than those used for sales and order-status information. We had 17 databases that didn’t always communicate easily with each other.

We decided against creating a program from scratch to capture customer information. That would have been years in the making. A lot of software would have had to be written and tested to re-create the basic features of an existing system.

Some features we wanted—such as enriching customer records with a Dun & Bradstreet credit analysis of each company, or automatically rooting out duplicate records—were nice to have and technically elegant, but they did not fundamentally change the way our business functions operated, nor did they improve our ability to gather information from our systems. So we opted not to pursue them.

We pulled together the best Cisco talent—architects, engineers, project managers and analysts. The team considered off-the-shelf products that fit our business. We looked at database applications we already owned from Siebel Systems, Oracle and Epiphany, and gave the team 30 days to identify the product with a data model that most closely matched the functionality we needed. We also wanted something that easily integrated with our other applications.

Oracle won out. We were already using Oracle for manufacturing, finance and customer service applications. Within 10 months were able to get a “single source of truth” about our customers.

The project wasn’t about having the biggest database or best customer information system on the planet; it was about being more efficient with our core business and understanding our customers’ needs.

I don’t want to diminish the effort behind this project. It was no small task. A team worked all hours to extract current information from multiple databases, clean it up and put it into the new format.

In one situation, more than 5,000 objects had to be converted to the new format, touching 300 million-plus rows of data. We primarily had overlapping data that led to multiple entries for a given piece of information. Before the conversion, one customer could show up in our databases as multiple companies. Simple input issues such as periods or abbreviations would make older databases think it was a new entry.

So we took records from the various databases—most of which had slightly different definitions of the same customer—corrected all the data to conform to the new definitions and then incorporated it into the new Oracle database.

What benefits are we seeing? We are beginning to understand our customers’ needs better and can determine which ones have a greater likelihood of buying another Cisco product or service. For example, certain vertical customers— such as those in health care, financial services or consumer products—can demonstrate higher probabilities of purchasing an Internet protocol telephony system six months after buying a network storage or security product.

Customers can now be broken into segments based on their purchasing behaviors. We have determined that some have a significantly higher probability than others of needing service. We also know who is using our products and how, so we can provide software patches or offer replacement products before a customer has a problem and logs a complaint with our technical assistance center.

Cisco can be proactive to its customers’ needs rather than reactive.

–Written with Anna Maria Virzi