International Technology Group 17
Costs of downtime were calculated based on estimates of reduced downtime under AVP arrangements for
problems caused by outages (typically corresponding to the IBM PMR classification of Severity 1) and,
in some cases, by severe impairment of application functionality (typically corresponding to Severity 2).
Examples of such estimates are that time to resolution for a specific problem would have been six hours if
IBM Technical Support only was employed, and two hours if AVP procedures were activated (i.e.,
downtime was reduced by four hours); and that time to resolution for a different problem would have
been 12 and 5 days respectively (i.e., downtime was reduced by 168 hours – the application in question
operated on a 24x7 basis).
Estimates were based on analysis of PMR records over a 12-month period for organizations that
contributed to this report and on customer input, industry experiences and values developed by the
authors of this report.
Two types of costs of downtime calculation were developed:
1. End-user productivity loss. Numbers of hours of downtime were multiplied by numbers of end
users affected, and by an application-specific value (typically 0.05 to 0.3) for reduced end-user
productivity while the application was unavailable or severely impaired. Allowance was also
made for the percent of users who would normally be active during this period.
The resulting value for total hours or days of lost end user productivity was then multiplied by
average end-user cost per hour based on average compensation (salary, bonus, benefits and other
items as appropriate) divided by numbers of hours worked per year.
Average compensation was calculated based on industry data for occupational categories (e.g.,
managers, executives, professionals, front-line employees, administrative employees, customer
service personnel) and industries (e.g., financial services, manufacturing, government).
An example of a lost end-user productivity calculation – in this case for an outage affecting a
Lotus Domino e-mail server – was as follows:
Six hours reduced downtime/impairment x 500 users (the average number of users
per server in this installation) x 0.75 active x 0.3 reduced productivity for x $53.17
average compensation per hour = $35,890.
2. Business calculation. Numbers of hours of downtime were multiplied by application-specific
business values for specific applications.
For example, for a retail Web site supported by DB2, calculations were based on average
numbers of visitors per hour, and average sales per visit. Using industry values, allowance was
made for lost sales that could be expected if visitors were unable to obtain information, place
orders or perform other tasks online; e.g., three percent of frustrated visitors would decide not to
purchase or purchase from another supplier.
In a number of installations, IM, Tivoli and/or WebSphere applications supported supply chain,
logistics and distribution operations. In these cases, hours of downtime were multiplied by
organization-specific values for average cost of supply chain disruption per hour. Industry and
ITG-developed metrics were employed to calculate these values.
Comparable organization-specific calculations were conducted for business intelligence, case
management, customer service, directory, financial, funds transfer, human resources,
infrastructure, marketing and sales, policy management, procurement, research, security,
transaction processing and other applications.