You are on page 1of 4

How to measure and improve AMS throughput

Executive summary Today, measuring IT metrics is very fashionable, yet, how do we make sure we determine the right metrics, when the customer is only concerned about receiving their service on time? Program managers can also suffer from a variation of this resource capacity problem. All too often, program managers can take all the conventional steps of ensuring good project management and yet forget that all the pieces of a program are not created equal. The easiest solution to this problem is to agree on some objective method for ensuring that all the business priorities are represented in the backlog. Adopting an approach that allows for requests to be classified into groupings similar to those used in the scorecard (financial, process, customer) tends to mitigate some of the charges of unfairness. Manufacturing techniques, such as the queuing theory, have found limited acceptance in IT for many reasons. Prioritizing IT projects is generally more difficult because not all projects are linked to revenue targets, and there is a belief that consciously setting aside money for cost overruns at the management level will be regarded as abomination by the finance department

Copyright 2007-2008. All rights reserved. Softtek Integration Systems, Inc.

How to measure and improve AMS throughput Key Findings When trying to prioritize projects, unclear value cases and subjective justifications create "parking lot" that make it difficult for managers to achieve a correct number of deliverables with the resources available. Limited availability of resources can cause projects to get backlogged. A project that assumes utilization of 100% of the staff without regard to skills or project timing will always end up with fewer completed projects than initially planned. The term on time implies that a date must be met. Bringing a project in on time requires meeting a series of dates because missed dates have a cascading effect throughout a project.

Key practices Projects Pipeline prioritization A collection of project practice that are developed and executed across a management domain; and that are consolidated into a single view of overall value and risks Practice Requirements Level 1 Nonexistent Informal Level 2 Initial Standard criteria Level 3 Standard Centralized & controlled Decisions made by type of work, standard estimation process Resources are allocated by detailed work assignments Active requirements are centralized Earned valued performed by individuals Level 4 Manage Clear measurements of benefits Historical measurements of key indicators Program estimations are balanced due to the statistics Detail data agreed cost, effort, scheduling, benefits Earned valued performed by groups Level 5 Optimize Delivering improve Active improvement to business Result of efficiency benchmarks Focus in continues improvements Continuous improvement strategy. Benefits are tangible by key indicators Detecting missing opportunities

Prioritization

No strategic alignment

Isolate criteria based on workload Resources are planned

Resources

Isolated process

Governance

Ad hoc process

Basic process

Performance measurements

Ad hoc process

Evaluation by effort

Risk

Ad hoc process

List of individual risk

Sensitive analysis by business

Focus on risk assessment performed by groups

Copyright 2007-2008. All rights reserved. Softtek Integration Systems, Inc.

How to measure and improve AMS throughput Increasing the Throughput of the Projects Pipeline Customer expectations Driving to fulfillment of expectations and delivery of the full scope as defined and understood is the primary function of project management. Problems arise when the setting and managing of expectations and scope are overly informal or conversely, overly formal. From our perspective, EVERY request requires the goals and objectives be written so that everyone is clear. The objectives must be mutually exclusive and strive to be a balance between being concise and comprehensive.

Classify the type of work This is a pretty good place to start. Requirements should not be treated the same so looking at request size makes sense. Segregate the type of request in order to process breaking down the work into type request group. Use segmentation to identify the different types of requests, in order to measure accurately by type and start looking into the variance.

Transform the practical problem into a statistical problem How can you measure the capacity of a team when every feature is different? Capacity is not an absolute number and that it is expected to have a spread of values. It is important to understand the mean and the variation. Capture each and all of individual transactions. Find a way of measuring them in terms that are meaningful to the outside - at the output but preferably at the input. To measure capacity, observe the throughput of value units over a period of time (e.g. 6 months) and sample the capacity for each month. The request types vary in scope, nature and size. However, if you collect data you can demonstrate that there is such a thing as an "average request" and that the capacity could be defined in terms of a mean and a reasonable minimum and maximum per month. Choose statistical diagnostics metrics to drive improvement It is critical that decision makers identify and choose no more than one metric that allows them to evaluate, rate, and compare different solution alternatives during similar time frames of measurement. Subsequent metrics may be chosen to address a different problem once the first problem has been solved. You have to start somewhere. Kill the variance first then shift the median. Do NOT try to do both at the same time

Copyright 2007-2008. All rights reserved. Softtek Integration Systems, Inc.

How to measure and improve AMS throughput Critical to Quality = Customer Expectation = Zero variance of days for every request Metric = Output versus Input = Days Early / Late versus Request

?
96 85 74 Y = Lifetime (Hrs)

Flashlight

During different intervals of time, observe the differences in the average times for the requests Prepared by: Gustavo Acosta Six Sigma Master Black Belt Application Maintenance & Support Global Practice Manager For further information please contact Raul Morales Business Relationship Manager Softtek p. +1 404.783.1539 e. raul.morales@softtek.com www.softtek.com

Copyright 2007-2008. All rights reserved. Softtek Integration Systems, Inc.

You might also like