Testing of software ranges from no testing to a full Software Quality Assurance (SQA) group. There are costs associated with each method along that spectrum that are employed.
The cost of not testing can be as extreme as incorrect calculations, lost data or lost revenue. Users are not usually trained in software testing so they are inefficient at determining and reporting issues. Assuming that no actual harm is being done by running the untested software, the best that happens is unproductive users as they struggle to do their jobs.
Automated testing such as unit testing and acceptance testing adds cost and time to the development cycle. Thorough manual testing has a similar effect. For these reasons it is tempting to eliminate testing to “save money”. Sometimes the testing is done in an ad hoc manner, randomly poking around in the application.
At the end of the spectrum with no testing, the costs are almost totally hidden in the overhead of daily business operation. Of course if orders are lost or prices are calculated wrong or some other very noticeable issue arises then the costs might become clearer. All software will be tested. Some applications will be tested by a test plan and others will be tested by trial and error.
Managing the costs up front where they are more visible is easier than waiting for them to appear later. Is it really cheaper to eliminate testing from the development cycle?
Recent Comments