Automation on a Thin Dime

Few would associate the cable industry with sophisticated process control techniques. Cable is challenged to deliver Internet protocol (IP) service automation, much less the continuous improvement linked to, say, the metrics and methodologies of Six Sigma. But improve it must. The only question is how. Our recommendation is to create an operation data store (ODS). Within that framework, we found that a small team on a tight budget could develop practical applications that led to reduced data errors and increased self-service transactions. The strategy is to gain cost-effective access to customer data and then evolve gradually into more automated processes. To get started, an operator first needs to face up to its legacy of mainframes and silos. The legacy Traditionally, independent operators outsourced front/back office information technology (IT) to cable billing systems, whose mainframe solutions allowed each region to customize its features. There was no need to develop internal IT expertise, and small teams working intimately together were able to deliver services efficiently without documenting process. Those 25-year-old legacy systems are now challenged to meet our customer needs, especially in complex IP-based services. The rise of internal silos (see Figure 1) poses the greatest challenge for customer service automation because automation requires known processes and access to relatively clean data. But the opportunity for success lies within those silos, namely by way of global reporting and loosely coupled automation. The strategy is to integrate the data silos and implement automated reporting; and discover undocumented processes and deliver continuous improvement through iterative development. Whereas the “waterfall” method (requirements analysis, design, implementation, testing/validation, integration and maintenance) works for telcos because of their historical focus on process and documentation, cable cannot easily automate using the telco or Big 5 consulting firm model because of variance in our undocumented processes and lack of legacy configuration control in our billing system. In our favor, cable has a “can do” attitude, Yankee ingenuity and a willingness to look at things anew. Getting it done Operators need to be clear on software requirements, aware of cost/performance tradeoffs and capable of building an ODS. Nailing the requirements is an underestimated first step; the operator must know and then manage automation requirements to maximize ROI from this project. It��s useful to have all participants take the customer perspective; then working across silos to integrate feature requests becomes easier. Because automation will create change regardless, this is also a good time to explore process and workflow alternatives. Another key is keeping team members aware of the exponential cost and performance implications of more features in a solution. As features increase, development and maintenance costs rise, and performance and reliability decline. (See Figure 2.) Cable can deliver value and reliability without the complexity of a telco. The mantra should be “keep it simple.” Finally, the team needs to be skilled in the database, messaging, and extraction, translation and load (ETL) functionality that comprise the technical framework of an automation solution. Common off-the-shelf software worked in our case. Using these components, we created an ODS that gathers data from billing, network and IP service provisioning. Master data elements are extracted with varying latency depending on several factors. Are they available? How necessary are they? And, most importantly, what is the cost of extraction? For example, overnight information from the billing system is relatively cheap, while zero latency read/insert/update is expensive. The network data, because of CableLabs�� DOCSIS standards, is easy to gather. The presentation layer for the ODS is a Web interface. Design priorities were simplicity, speed and ease of use. The interfaces focus on tasks that deliver automation value, the so-called 90 percent solution. If more complex issues arise, the underlying control systems are still available. Lessons learned Implementing this ODS (see Figure 3) over the past 12 months resulted in several lessons, corresponding to the phases of the project. Measure twice, cut once. Measurement reporting is the key to success. Data reports help you discover the availability and accuracy of data elements. For instance, how often is an element filled in? Does it vary in accuracy in each region? Process discovery reports let you validate the process statements made by each group. The reporting exceptions captured are the key to discovering undocumented processes. Existing reports, logic and provisioning logs have proved very useful to define or discover process. Answering these types of questions was made possible by the ODS��s easy access to all data sets, from engineering to customer data billing to provisioning logs. The operational principle is to gather the data once, then reuse it many times. Automate gradually. It is possible to implement simple, incremental solutions that reduce “swivel-chair” data errors, clean up data and increase rates of Web self-service and automated provisioning. In our case, the benefits have translated to a 63 percent reduction in data errors, a cleanup of orphaned email and modems, and email Web self-service transactions that now exceed those involving technical services reps (TSRs). Measure and improve. Discover how the application creates value and ROI; look for ways to improve it. Here are some questions indicating how an operator could use an ODS to improve an email self-service application: – How many emails were provisioned via Web self-service? – Is there an increase in email use because of self-service? – How many emails were provisioned during off hours? – Was there a reduction in calls by customers requesting email addresses? – Can we measure a customer-perceived increase in service quality? – What is the churn in customers who have email with you vs. high-speed data customers who do not use your email system? The payoff to creating an ODS is a virtual integration that standardizes processes, promotes automation and in-house collaboration, maximizes economies of scale, and gives operators a way to measure process and performance for broadband services and products. It may not approach the Six Sigma goal of 3.4 defects per million opportunities (DMPO), but it is a step forward in customer service at an effective price and represents the first step toward continuous improvement. This shared repository with engineering, customer usage and business metrics gives operators a practical way to turn data into information and thereby improve operations. Tom McCarthy is director, engineering, at Millennium Digital Media. Reach him at [email protected].