Performance Improvement 101

Many companies use benchmarking as a management technique to investigate and compare performance against internal standards and external peer organizations. As such, benchmarking provides useful information regarding a companys relative standing.

A high-level benchmarking exercise is limited, however, as it merely scratches the surface of performance and yields few insights into underlying causes, or into what can be done to improve.

Root Cause Analysis

Effective performance analysis and improvement involves collecting data (quantitative and qualitative) in the target area, and establishing a hierarchy of metrics to represent the operation or process. A high-level analysis is useful to an extent, as it can show how an organization performs and, in a comparative context, how it performs relative to its peers.

The chart above compares network hardware unit costs for Acme (top number) against network hardware costs achieved by top-performing companies (lower number). At first glance, Acme seems to be performing well, since costs are comparable for the high-level measure of network hardware costs ($74 vs. $73).

However, an anomaly appears at the sub-process level, specifically in the area of Acmes voice network hardware costs. But the impact of this anomaly on overall performance is obscured by lower-than-average costs in data network hardware. Moreover, at this level, while the differences in voice and data network hardware costs are indicated, the reasons behind those cost differences are not apparent.

Drilling down further (chart below), we see that PBX costs are somewhat higher for Acme. But to understand the real source of the issue, its necessary to go even further down the metrics hierarchyto large PBX maintenance costs, where Acmes costs are more than double the Reference Groups.

Once this root cause of Acmes high voice network hardware costs is identified, corrective actionand a significant improvement in operational efficiencybecomes possible.

But what specific actions should be taken to lower Acmes high costs in this area?

Compass analyses indicate that high PBX maintenance costs commonly result from one (or more) of the following factors:

• Non-competitive service procurement. In this case, the client organization should renegotiate its maintenance services contract, either through a formal RFP process or through an informal multivendor approach.
• Unnecessary maintenance of non-operating PBX components. Unused port and trunk cards should be removed from the PBXs and consolidated to minimize the number of card shelves; cards that have been removed can then be used as maintenance spares or to support growth in other locations using similar PBXs.
• Unnecessary maintenance of spare PBXs. Leading practices indicate that maintenance services are not necessary for spare PBX components.
• Inappropriate service coverage or response time environments. User organizations should review their PBX service coverage needs (24×7 vs. 12×6 vs. business day, etc.) and buy the level of coverage necessary. Similarly, response time for outage coverage should meet business requirements: Is a one-hour response necessary for a port outage? Or is same-day coverage adequate?

Applying the Methodology Across Service Towers

Similar analyses can be applied to sub-process activities in other IT infrastructure towers to formulate specific actions to improve cost efficiency, productivity and quality. In a desktop environment, for example, the high-level metric of cost per desktop may not reveal potential improvement opportunities that exist further down the metrics hierarchy, in an area such as support personnel per desktop. In a mainframe environment, the root causes of performance gaps in the high-level measure of cost per MIPS might be found in areas such as storage utilization or acquisition practices. In an application development environment, low productivity might reflect an inadequate system of activity reporting, resulting in limited knowledge sharing.

Its important to remember that cost reduction in and of itself is not necessarily the outcome or the goal of a detailed performance analysis. In many cases, the analysis can reveal that a lack of investment in technology, personnel or processes relative to top performers results in sub-par performance. Compass data shows, for example, that leading service desks have invested in self-help and online tools to boost agent productivity and increase user satisfaction.

The objective of performance analysis, moreover, should never be to focus on which organization performs best or worst. Rather, the aim is to identify best practices and to understand their impact on performance, and to incorporate those practices to optimize benefits throughout the organization.

Performance analysis can improve IT operational efficiency and quality in a variety of ways. An effective measurement system provides a foundation for better decision-making and realization of tangible results, both in terms of operational performance and ITs contribution to the business.

Carl Pitasi is a Compass senior consultant based in New York City.