Expert Judgment (Individual)
Many organizations don't have processes or tools for estimating. Formal estimation methods or project experience databases aren't necessarily required to provide a qualitative effort estimation. Expert judgment is the most commonly used estimation approach in practice. Expert Judgment is based upon a specific set of criteria and/or expertise from a knowledge or product area, a particular industry etc. This knowledge base is provided by a project team member, team leader or external group/person. Often expert judgment requires an expertise outside the project team, so external groups are consulted.
Description
This method involves consulting one or more experts. The experts provide estimates using their own methods and experience. When you use Expert Judgment you have to ensure that it is effective. Lederer and Parsad evaluated that estimates provided by people who aren't actually doing the work are less accurate. So, to guarantee that good estimates are provided you have to separate tasks into smaller ones and assign their estimations to people who have specific knowledge in the tasks area. When making estimations developers, testers and managers tend to de-emphasize tasks that they don't understand. Tasks shall be decomposed that will require no more than 2 days of effort, otherwise they could contain too many unforeseen extra effort.
How To
It happens all too often that expert judgment is overoptimistic. Effort estimates are often created by bidding, best case assumptions or most likely effort assumptions. This might happen unconscious, if - for example - a project deadline exists. Therefore don't let the person estimating know such information like price-to-win assumptions and separate planning and bidding from effort estimation. A way to increase quality of estimates is to ask for justification of cost estimates.A technical skill level can be a poor indicator of estimations accuracy. Hence estimation from more than one expert will mostly deliver a more robust and accurate estimation considering the estimators used different estimation principles and knowledge. Besides estimates are more applicable if experts are used that have experience from similar projects. It has to be considered whether an expert has expertise in how much effort a task will take not only how to perform a task.
To train a team member to provide adequate estimates, it is often useful to let them additionally create Best Case and Worst Case estimations for a task. Afterwards compare the estimates with the originally single-point result. This exercise will show the estimator that single-point estimates tend to be akin to Best Case estimates. Besides it will assure that worst case scenarios will be considered as well. It is also useful to establish checklists. They help to improve that everything is considered. A estimator shall also keep track of the actual results (time spent on the task) and compare the results to the estimated work. The estimator shall try to understand what went wrong, what went right, what was overlooked and how to avoid these mistakes in the future. The key principle is to set up a feedback loop. This can also be done by a public estimation review.
The following two approaches briefly describe how to break up a project into small components.
Top Down
First an overview of the system is formulated. Afterwards subsystems of the first level are drafted. Each subsystem is then decomposed into smaller subsystems. The process is repeated until the whole system is reduced to the base elements.
Bottom up
The bottom up approach follows the work breakdown structure (WBS), see [Decomposition by Workbreakdown Structure] idea and is used for projects where no appropriate historical data is available. The project is split into component tasks. These component tasks are again broken into subcomponents. The process is repeated until the work effort for each subcomponents is not more than a couple of weeks per person.