Justification of Potential Percentage
Time Saving Figures

How much could you save on your software testing?

The cost saving calculator contains a number of percentage saving figures which have been based on findings during actual projects, with a further conservative adjustment. When completing the form you obviously want to be happy that the times, costs and percentages used are appropriate. The purpose of this page is to describe some of the real-life situations we have found with existing customers and the actual percentage time savings they have achieved.

Unit Testing

Castrol UK. This is a Unit testing/System testing situation. Castrol's well quoted and documented story is about intensive tests performed totally automatically on a key batch program that calculates prices. They have gradually built one of the most thorough test cases we have ever seen, which they add to with each change to the program, business or with new insight. It automatically calls the program with every scenario they ever imagined and checks the output automatically. It is a very critical part of their system and the test they have built takes 5 hours to run (not a very fast CISC machine). But, to test it as much as they could manually (and they would never achieve the same in depth coverage) they estimate would take around 12 days. Again let's be generous and say they have reduced the time from 12 days to one day. That's still a time saving of 92%. But, that is not what they care about. What is really important to them is that they know that this part of the system has been tested as well as it possibly can and that the risk of an error going into production is minimal. Not everyone will achieve this sort of saving, hence we have used only a very conservative 50% in the spreadsheet.

Test Data Creation and Management

Fidelity Investments bought TestBench initially to save time in the extraction of data. A senior person was spending about 2.5 days per week creating environments for various project teams and tests. She understood the database so the task required her knowledge and experience. She set up the Data Cases she wanted to perform the extractions, and thereafter spent only 0.5 days per week on the task. A straight 80% saving, but with only 70% used in the spreadsheet.

System Testing

Dresdner Kleinwort Wasserstein Bank in London This was more of a system testing project. DKW were ready to test, they already had a plan which they had estimated was going to take then four weeks to do manually. They added a week to the plan to get TestBench up and running. During the first week (planned for set up) we did 3 weeks worth of testing. In percentage terms it would be fair to say they did 5 weeks work in less than 2. Hence a cautious 60% saving in the spreadsheet. Click here for press coverage of this.

Regression and User Acceptance Testing

Royal Bank Invoice Finance. Their prime focus is User Acceptance testing. They had a test plan that took 6 weeks (and 1.5 persons) to execute and check - assuming it all went well. They put the same thing into TestBench for iSeries and it took one week to perform the same tests - with 1 person. This appeared in the press as a case study and we quoted an 83% saving in time. And that does not take account the reduction from 1.5 to 1 persons to complete the task. To be totally accurate, we should point out that now they spend about 2 weeks testing each new release, which is a little bit longer - but the reason is that they now do more testing than they did before. Their goal was not just to save time and effort, but to reduce risk to the business. They did both. Hence we believe that to suggest a 60% saving in User Acceptance Testing or Regression Testing is not unrealistic.

Rework

Estimating the reduction in rework is harder, but if the number of production incidents can be halved (and that is easily achievable) then, together with the extra supporting information and the improvements in the re-testing, a 70% saving would not be over optimistic.

Back to Top^