Estimation: Experience of the person doing the estimate

This is the single most important factor – but also the hardest bit to gain. It’s a classic Catch 22, you need to be experienced but how do you gain this without making mistakes? I’m not here to solve that one just to let you know this is vital. If pushed for advice best bet if you don’t have the skills already from past projects is to do prototyping first and/or get experienced external help.

If the person doing the estimation missies out common tasks (e.g. code reviews, cross browser testing, …) or does not understand the capabilities of the tools and platform they are using this is cause for great concern.

I’ll get onto talking about techniques for estimation (e.g Top down or bottom up) shortly. I’ll also chat about tools you can use to assist you and several common tasks. These will help someone to estimate a project well and work out the core activities.

However, what they [techniques and lists of common activities] won’t help with is that you do _need_ a degree of platform experience - if you asked someone to cost up a project to build an on-line portal for users then it is vital that the estimator recognizes the strengths of the technology that will be used i.e. ASP .NET 2.0 (would for many scenarios) be much easier than using another technology e.g. ASP .NET 1.0, ASP 3.0 or JSP/ Servlets.

As an aside, once the project is started detailed requirements capture I’d also suggest that it is pretty important that the analyst has a good understanding of the features of the toolset as well – that way they will realize the costs of simple changes to functionality that move outside of the core platform support e.g. ASP .NET Validation.

 

I love being able to write these 'in a perfect world' posts ;)

Comments

• Anonymous
  March 21, 2005
  First fly that typically enters the ointment: Marketing Person/Soulless Suited One had already provided an estimate based on what they believe will make certain people happy.
  At which point when you are asked for an estimate it goes like this:
  Manager: How long will it take and how much will it cost?
  You: [reply with result of doing homework based on experience etc.]
  Manager: Wrong, guess again.
  
  At which point we're heading towards some more Quantum Project Management (which relies on the task being worked on simultaneously in an infinite number of parallel universes).
  
  Sorry, I must try not to let cynicism overwhelm me.
  You're doing well, I look forward to the next instalment with interest.
  
• Anonymous
  March 21, 2005
  The comment has been removed