July 1st, 2014 by Adam Sandman
I have often been presented with the argument that QA is not cost effective. It seems like a throwback to years past, but it happens all too often. With the advent of Agile, and other rapid development cycle methodologies, the mindset of "we will fix it once it is reported" has made a comeback. I know the MVP process, and it does include a specification, and can therefore be tested.
Preaching to the choir here, but this is complete hogwash. We need a new primer on not the cost of QA, but the cost of skipping QA.
Credibility
The first cost is to your credibility. Sure your customer might not find the defect you did not test for, but if they do, will they remain a customer. There are plenty of software providers, and the difference a defect makes may determine whether you maintain your current customers. If you lose your current customers, what is going to happen to future customers? Who will they call, or see on the internet message boards? That’s right, your dissatisfied previous customers. If it is internal, the cost in credibility may have an effect on your promote ability, or outsourcing of the department.
Usability
Next up is usability. Sure the manual says not to do that keystroke, but guess what? They are going to do just that. Affecting the usability of an application because we are afraid of defects that we did not test for is hobbling the application. Users do not like hobbled applications and will look for more full featured solutions
If we don't find, and subsequently fix, the defect now, how do we know it won’t be in future revs? The easiest time to fix a defect is when the defect is young, before it gets its roots wrapped around other features and makes the fix take ten times longer. When we try to fix deep problems, we often create additional issues that will require additional cycles. Software is complex, squash the bugs while you can.
Escalating Costs to Fix
As I referred to, a fix late in life takes more cycles. More cycles cost more money. Saving a week on a release, skipping the qa cycle, may seem like a good thing, but it will cost in the long run, 10x the effort to fix an issue is 10x the money. Budgets are tight, but this line pays for itself in avoiding an expensive future fix.
Reputation
As a software vendor, internal or external, our reputation is intrinsically tied to our customers’ reputation. Do you want to be the provider who developed the software for a company that crashes or causes some harm? Even if it was directly to spec, bad software leaves a stench, and that stench kills companies.
In Summary
I do not care if you are agile. I do not care if you have a deadline. Do not skip the QA cycle. If you need a shorter time frame, make sure you have the tools that allow for efficiency and let your people save you money in the long run.