[/B][/QUOTE]
Here is a statistic that will back up your claim. Taken from the book “Automated Software Testing,” 5th printing, page 8.
" …Refer to Table 1.1, which outlines the cost savings of error detection through the various stages of the development life cycle.
Error Removal Cost Multiplies over System Development Life Cycle ( B Littlewood, “Software Reliability: Achievement and Assessment.” Blackwell Scientific Publications, 1987.)
Phase Cost
Definition $ 1
High-Level Design $ 2
Low-Level Design $ 5
Code $ 10
Unit Test $ 15
Integration Test $ 22
System Test $ 50
Post-Delivery $ 100+
Table 1.1 Prevention is Cheaper Than Cure "
In addition to the various statistics available that show the fact, that it is cheaper to fix a defect if it is found earlier in the development life-cycle, here are some basic “common sense” ideas you can use to state your case (and come up with a cost estimate):
 How does the production software get to your customer? Do you ship CDs? If yes, what is the cost of having to re-burn and re-ship the CDs?
 If you don’t ship CDs, who installs the software? Someone from your company? How many people are involved in installing it at the customer site? What is the cost of having to re-install the software at the customers site? What is the cost of the customer having to re-install the software at their site? What is the cost of tech support having to explain the re-installation to the customer?
 If major showstoppers are discovered in production, what about customer satisfaction? How many customers have canceled the product, because of those defects? What is the cost of loosing these customers? What is the cost of the dissatisfaction of this customer – if they didn’t cancel (not as easy to quantify)? How many customers don’t re-subscribe, because of this dissatisfaction?
 How many customers are calling in about the showstopper? How much time does tech support spend, having to give the customer advise explaining the problem to them? Estimate and add up the extra hours that tech support has to spend on customer support.
 How many people are involved in the software development lifecycle? If a defect is found in production, how many people are required to work on tracking down the problem (i.e. tech support, testing team, etc), analyze and implement the rework required, re-specification, re-design, re-test, etc? Estimate the time/$$ of those people involved (not to mention the cost of frustration, turn-over rate, etc).
These are just some ideas that quickly come to mind. There are additional statistics available that show, how much cheaper it is to fix a defect earlier in the development life-cycle.
------------------
Elfriede Dustin
Author (with Rashka, Paul)of book "Automated Software Testing", July ‘99
Author (with Rashka, McDiarmid) of book "Quality Web Systems: Performance, Security & Usability", August ‘01
www.qualitywebsys.com
Please click the following link to reply:
http://www.qaforums.com/Forum15/HTML/002240.html