Discussion Topics

« Return to Sizing Methods

Industry standards for typical time spent on various project phases

Hi,
Need some help! I am looking for some industry standards / metrics for typical time spent on various project phases.

We are following iterative development approach. My project is web based enterprise application (complex calculation engine) with clear stated stakeholder expectations around requirements. We are planning to use Web 2.0 / J2EE / Oracle.

I am looking for typical time (%) I should be allocating for following phases:

Project Management / Planning
Requirement Analysis
Design
Coding
Testing (system test, integration test, UAT)
Rollout / Transition

Any ideas / experience / research paper would be much appreciated.

Thanks
Kashyap

Replies to this Topic

Download an eval copy of Cost Xpert (www.costxpert.com). The estimation tool has % values for a lot of project types/domains (web, iterative etc) readily available under the 'maintain' menu.

Let me know if you have problem accessing the numbers.

Gerrit

Edited Sat, Feb 2, 2008 10:49 PM

Also, the QSM Almanac (http://www.qsm.com/learn_morec.htm) is a good source.

 

From their website:

"# Get a snapshot view of 564 IT projects
# Learn the characteristics of the industry's best, worst in class performers
# Empirically analyze interactions between time, effort, size, productivity, and software reuse
# Get in depth analysis of industry trends drawn from nearly 5000 IT projects"

Edited Sat, Feb 2, 2008 11:01 PM

While we're trying to use quantitative approaches, I think we also need to be precise and accurate with respect to the language we use to discuss the subject. Hence, there can be no 'standards' for the relative percentage of effort (or duration, or cost, etc) expended on the work to produce different kinds of work-product. We may be able to identify 'industry averages', perhaps… but we should not fall into the trap of thinking of these as 'standards'.

Also, what you mean by 'iterative' must be very different from my understanding of this term. In my experience, iterative projects aim to deliver work-products in each of a series of time-boxed periods ('iterations' or 'sprints' if you will) either directly for immediate use or at least to into the user acceptance test process. In such an approach, the relative effort allocated (and expended) for each 'phase' is the same as asking 'how much effort should I allocate to each successive transformation in the product life-cycle'. That is, requirements to design, design to unit-tested code, code to acceptable releasable product. In an iterative project you can easily use the data from each successive sprint to (re)calibrate your initial allocation of [ effort | time | resource | cost ] … so, by reducing the duration of each iteration, you can make your problem disappear!

 

Dear Kashyap,

You may wish to read the following:

 http://www.stsc.hill.af.mil/resources/tech_docs/gsam3/chap13.pdf

http://www.cis.um.edu.mt/~abut/

http://www.oit.state.pa.us/oaoit/lib/oaoit/enterprise_projects/example-performance_metrics_for_software_development_v1_4-26-05.doc

The results of your project will depend upon many things. 

How well us the scope of your effort defined? How details are the requirements. How much expertise does the tea, have with the techniology and the problem domian?  

Never the less, I would recommend an iterative or spiral approach:

http://infolab.stanford.edu/~burback/watersluice/node53.html

If using a modeling tool:

http://www.telelogic.com/services/process_improvement/harmony_itsw.cfm

Software development using a waterfall approach is not recommended. 

As far as project estimation goes, the better job of capturing metrics, the better job of process improvement and estimation you will do in the future. For autoamted code based metrics try:

http://code.google.com/p/hackystat/

Dennis

Post Reply

You must be logged in and a member of this Groupsite in order to post a reply to this topic.
To post a reply, contact your group manager(s) Join this Groupsite