By now, we know that the first objective of any IT project portfolio management (PPM) practice must be to determine which projects are the “right” projects for a company to invest in.
To accomplish this objective, a successful PPM practice is comprised of several lower-level evaluation and analytical disciplines. One of which is an effective project estimation method.
A company’s IT department is required to provide estimates to the business units whenever the business units request services from IT. These estimates occur several times in the life of a project.
Specifically, project estimation should occur at four phases in the project lifecycle as follows:
Collectively, these estimates should work in a cumulative fashion, building on one another and providing more accurate estimates at each phase.
Picture a funnel with the large end on the left-hand side and the small end on the right-hand side. If you were to place a stick down the center of the funnel you would have space between the stick and the edges of the funnel both above and below the stick.
Imagine the stick (going straight through the center) as your estimate and the space above and below the stick as the wiggle-room associated to the estimate. The edges of the funnel then, represent the variance levels applied to the wiggle-room.
At the left-hand side (i.e., the widest opening) the wiggle-room is largest. This is synonymous to the order of magnitude estimate that comes out of the screening phase, which has the largest tolerance for error.
The right-hand side opening of the funnel (i.e., the smallest opening) is synonymous to the implementation estimate that comes out of the technical design phase.
To help make things clearer, let me summarize the estimate types and appropriate variance levels for each:
Screening/Order of Magnitude Estimate: This high-level estimate, sometimes called a “ballpark estimate” is based on high-level objectives and major deliverables usually described in a paragraph or two supplied by the business on a project request form.
Most order-of-magnitude (OOM) estimates have a range of variance from –20% all the way to +80%. Like I said, lots of wiggle room.
Your IT department shouldn't invest too much time in creating these initial estimates and the customer shouldn't place too much confidence in the accuracy of the OOM estimate.
Unfortunately, a breakdown in expectations can easily occur when it comes to OOM estimates because the IT organization tries to make it as accurate as possible and provides estimates in numbers such as 3,244 project hours or $123,456. These kinds of numbers promote accuracy to the business customer and thus set expectations inappropriately.
Instead, all OOM estimated values should be rounded to appropriate levels to illustrate their ballpark-like accuracy. Using the previous examples, I would recommend that you round 3,244 to 3,500 and $123,456 to $125,000 or possible $150,000 to ensure the appropriate expectations are set.