Note 4: Service Assurance and Data Analysis

Posted: 23 September 2005

Service Assurance

Service assurance, like provisioning, is a network interfacing function. When network-related problems are identified, assurance sees that they are managed until fixed. Assurance should also include the proactive identification of otherwise unreported problems and their management up to resolution.

When problems occur, the biller's staff capture the problem details provided by the customer. Problems are logged and prioritised based on the staff available and the expertise required. Once allocated to staff, a problem's cause is identified and the appropriate fix applied. Some problems can be fixed remotely, whilst others may require the dispatch of network staff to the customer's location.

Any shared use of network staff between provisioning and assurance requires that the competing priorities of new orders and network problems be carefully managed (work force management). The different skill levels between staff, and dedicated teams for specific tasks (e.g. new connections) can cause complications. New connections can be relatively standard; problem resolution usually requires more experienced staff. More complex network technologies (e.g. telecommunications) may require a detailed understanding of a range of technology platforms.

Different industries (based on their networks) will require different levels of assurance activities. Billing of government charges based on a residential address' details may have assurance issues concerning the postal address for documents, who owns a property and its current value. These can be addressed from a central office location during business hours. Water networks will require registered plumbers with access to underground pipe information, equipment to replace faulty pipes, and details of where to turn the water off. These details must be available to a large, distributed workforce operating 24 hours a day from trucks and vans.

When multiple problems occur, assurance must identify whether there is a common cause. The identification of a common cause can sometimes reduce the overall resolution effort by targeting the underlying fault rather than each individual problem. However in other cases, each individual problem may still need to be tested and closed. Identifying an underlying fault can also identify the broader scope of an outage / issue and allow the identification of unreported impacts. Proactive resolution and notification can then provide a higher level of customer service.

Even more than ordering, assurance uses service level monitoring to identify when outstanding problems are in jeopardy of breaching service level agreements. The time allowed to resolve problems will normally be included in commercial contracts, or, in non-commercial circumstances, by minimum regulatory standards.

A premier level of 'service restoral' may be offered to customers as a 'value-add' product. Alternatively, essential service providers (e.g. hospital, doctor's surgery or police station), or emergency situations may require a premier service. Additional information on such situations can be retrieved from the billing system so that it can influence the priority with which a specific problem is addressed.

When service levels are breached, rebates and contractual credits may apply. When a problem is a customer's fault there may be 'time and materials' charges applied to recover the biller's costs. These internally generated, non-network charges are forwarded to billing for inclusion on the customer's bill.

Data Analysis

Data analysis combines data from different sources, validates its accuracy, transforms its format to one suitable for high-performance analysis processing (rather than transaction processing), and assembles it ready for use. For billers, data analysis can provide insights into customer, transaction and network information that are not easily obtained using regular database queries.

For example, data warehouses can consolidate information from across the biller's systems to enable integrated dissection of the biller's data. Queries can be performed by business staff to identify new sales opportunities, regional performance differences, clarify the effectiveness of different marketing campaigns and track long-term purchasing trends.

By analysing behaviour over a sufficiently long time period and for a large customer sample, a biller can construct predictive models of customers' propensity to perform future activities. Examples include the customers' likelihood to purchase (e.g. useful for sales and marketing), or change providers at the end of their contract period (i.e. churn). It can also be one of the data sources used in fraud detection.

Fraud Detection

Fraud detection analyses data to identify fraudulent or suspicious behaviour. The specific methods used to perform this analysis are often restricted to a select group to ensure the 'alarm's tripwires' remain hidden. Suspicious activity may or may not be criminal. A customer's change of behaviour may be based on a change of circumstances, and, just like a credit card company confirming an unusual purchase, once investigated can be dismissed from further consideration. The customer's new circumstances may present sales opportunities that can be passed to other business functions.

For high-value transactions, fraud detection may need to be performed in near real-time in order to limit the biller's financial exposure. Lower value transactions may be analysed after a brief delay using cheaper infrastructure than required for near real-time checking.

Marketing Support

Billing can be an information source used to support the biller's marketing initiatives. Billing's core databases can include information such as contact details, market segmentation information, and details of recent purchases. Billing's accurate information is important where regulator-mandated codes-of-conduct limit the frequency with which a customer can be contacted. Candidate targets for marketing campaigns may be identified by the data analysis function.

Tags: Billing, Service Assurance, Marketing, Fraud