cPrime

Estimating, an Art or a Science?

When asked how long something might take, do you think of a number and then double it, subtract your age and multiply by the number of letters in your pet's name? Is accurate project estimation an art or a science? What we’ll explain is that there is no definitive answer to this question. The most successful project estimation occurs when both art and science are blended together. This article will discuss estimation best practices along with tips and tricks, including approaches for top down and bottom up estimation. In order to draw conclusions we will first need to explain some approaches to estimation and some related concepts.

Note: There are many subtle iterations of both top down and bottom up estimation. For the purposes of this article, we’ll focus on the vanilla versions of both. Additionally, cost and schedule estimation can be intermixed or interchanged. In a fixed unit price environment, both cost and schedule are indelibly connected.

Analogous Project Estimation (Top Down)

Previous experience or historical project data will give you the ability to draw accurate estimates for new “analogous” projects. This is not always available to your organization but if you’ve been around a while, IT SHOULD. The value of this information should not be underestimated, it is critical to your business and PMO. If you’re currently not tracking actual project data, now would be a fantastic time to start.

How to Use: If Project A took 1,200 hours of total effort to complete, then it would stand to reason that Project B, if analogous , should also take around 1,200 hours of total effort. If the historical records of project A are extensive and accurate, this approach will be more scientific. The estimation team would need to focus on areas in which the projects vary to adjust estimates. Remember, the key here is accurate project estimating, so if you records are deficient, the pendulum swings to the artistic side of the equation.

Tips & Tricks: Keeping engineers tracking time to projects and project tasks is a great way to ensure that down the road you and your organization will have good estimates upon which to apply this methodology. KEEP WBS items specific enough to be useful to your engineers but don’t get crazy with it. Try to keep a WBS to the third level ONLY for tracking time and costs. Engineers hate tracking time and they hate overly complex time tracking mechanisms even more. Last, don’t use your best and most talented resource to tell you how long something might take, use your Mr. Average to estimate; it’s never fun to have the SME tell you it will take 2 hours when it takes the average team member 8.

Bottom Up Estimation

This approach gets its name because it works up from the bottom of the WBS; or the sum of all the project activity estimates. This is a much more analytical approach to estimation and requires a much more accurate set of project documentation, including but not excluding, customer/user requirements, functional /technical specifications and most importantly a complete WBS or Project Plan/Schedule.

How to use:

Start by breaking your project down into its smallest work components. Typically this will be done via a WBS.
Next, estimate the effort, duration, and cost of each. The artistic side of this piece is in knowing your team, their skill sets, how productive they are and how much fluff your engineers are likely to add, etc. (See Tips and Tricks.)
Aggregate them into a full estimate. This is pretty easy, if and ONLY IF, you did the first two steps accurately and completely.

Similar to Top Down estimation, the quality and quantity of useful, complete information will tell us how much “art” needs to be used in Bottom Up Estimation. If the scope, business/ functional requirements, and the project team skill sets are well defined, a project team can rely on primarily scientific approaches to “accurate estimation”. However, if these items are vague, such as a new team or vendor being used, the pendulum swings from primarily scientific estimation to a blend more dependent on artistic estimation.

Tips & Tricks: Know your team! There can be a wide variance in productivity from team member to team member so simply assigning any FTE with 40 hours of work effort to complete Task A is not enough. Also, be careful that you don’t get carried away with detail in your project plan or WBS. There is a point where the more detail you add, the more tendency there is to overestimate.

Why Estimates Are Wrong

We’ve covered the two main approaches of estimation. A commonality we discussed is that the more accurate historical data and project documentation is in combination with how well you know your team’s skills allows you to determine how much art will be required. Let’s take a moment to discuss some other things critical to accurate project estimation. Here are several ideas to consider and also help ideas to ensure more accurate estimation.

Understand Statistical Risk Analysis – How interested is your organization in gambling? Like great poker players, a great project manager should always know the odds on any hand of winning (aka project success). How do you go about this? Understand statistical risk analysis and the associated logic behind them; once you grasp these concepts, you will have an “Ah Ha!” moment and you’ll realize you’ve been setting out at a 30% likelihood of success or perhaps even worse!
Function Point Analysis – Determine how many functions or features you have, give them a level of complexity from low, medium or high and then using a formula suited to the type of project you are doing(Enhancement, Development etc), you come to a suitable estimate of resourcing. Of course, if you have limited data to put in, expect limited data out.
Monte Carlo Simulations – These simulations are best used when you “don’t know what you don’t know.” In order to estimate, you would first define all of the information you DO know and input into the system. The system then generates inputs randomly from the domain and performs computations. The results are aggregated and you find the “most likely” solution.
Deltek Risk+ – A comprehensive risk analysis tool that integrates seamlessly with Microsoft® Project to quantify the cost and schedule uncertainty associated with project plans. A great tool to help you with the Monte Carlo Simulations. http://www.deltek.com/products/riskplus/
Excel – There are addin’s for Microsoft Excel that serve the same purpose, perhaps just not as easy to use. http://www.palisade.com/risk/default.asp

Higher Complexity = Higher Risk = Higher Contingency

One of the fundamental mistakes we make as project managers is to assume that if the project is twice as complex, it will cost exactly twice as much. In practice, there are times when a project that is twice as complex might be as much as 4 or 5 times more costly.

Group Dynamics – As your teams become bigger and bigger, communication, collaboration, working styles, etc, all come into play to reduce project output. It’s similar to the “Law of Diminishing Returns”. According to this relationship, in a production system with fixed and variable inputs beyond some point, each additional unit of variable input yields less and less additional output. And the scary piece is that this is not a linear curve but potentially exponential. It has been suggested that a project 10 times as complex, might be 1000 times as costly!

As a result, the project manager needs to understand this relationship and appropriately plan during the estimation process. Appropriate worst case/best case scenarios should be used in your risk modeling, as well as appropriate contingency in your budget and schedule, to offset the increased risk component.

Conclusion – What Percent Science? What Percent Art?

Estimation techniques will always involve some guesswork and assumptions. However, those guesses and assumptions will be far better served if they are applied, as a last step, to a strong mathematical and statistical project estimation analysis. Also, be sure to remember that estimating is not something you only do at the beginning, you are constantly adjusting to changes in scope, more information coming in to hone in your estimations etc, they are not a single point in time calculation.

So, in general, here’s the rule common to both Top Down and Bottom Up Estimation.

If you estimating a project that doesn’t have any organizational history, is not well defined and documented or if you have new team members or vendors, etc. – Start with science, not exceeding your point of knowledge, and then apply plenty of art to come to an accurate estimate.
If you are working on a project with substantial documented history, and the project and team are well defined – You can rely primarily on science and apply art as a final “sniff” test.

So the key isn’t if Estimation is an Art or a Science, it is clearly dependent on both. The key is in the How, How Much and When do you trust your experience and gut when finalizing a project estimate?

Return to Top

Home :: Services :: Why cPrime :: Knowledge Center :: Global Responsibility :: Careers :: Contact