As I sit here waiting on yet another round of fixes to download, I had a few random thoughts about Quality Assurance (QA).
Perfection is the enemy.
In the QA world perfection comes at a price. I would argue that perfection comes at too great a price to ever achieve it, but whilst I would make that argument, I would concede that your pursuit of perfection should continue because it’s in that pursuit that you arrive at an equilibrium that is acceptable. For instance (purely arbitrary numbers), if 97% perfect output (i.e., no lost product) costs $1,000,000 to achieve, then 98% is going to cost $10,000,000 to achieve, and 99% $100,000,000 to achieve and 100% is going to cost $1,000,000,000 (or more) to achieve. In manufacturing, if the value of the lost product between 97% perfect output and 100% perfect output equals $5,000,000 then it obviously doesn’t make economic sense to pursue that extra bit of perfection (because it would cost you $1bn to recover that $5,000,000). In the service industry, think of this output as customer satisfaction. So, step one is establishing what is an acceptable, cost effective, level of output (in this case customer satisfaction). The higher the satisfaction level, the higher the cost. The company leadership must decide what that level is and then resource to achieve it. But once that level is established, you must be ruthless in your pursuit of it because it is a perishable commodity and decays over time, mostly due to complacency.
Beyond the costs, there are other variables that prevent perfection (human complacency, incompetence, lack of resources, etc.). An effective QA system though can address and reduce the impact of each of those. A QA team’s primary duty is to identify the shortfalls in the available system(s) and eliminate all possible opportunities for less than perfect outcomes. But, with all that, the QA Professional must ensure that the system for detecting, cataloging, fixing, and disseminating information about faults is as robust and foolproof as humanly possible. If every day they aren’t eliminating a possibility for a repeat fault, then that QA team is failing. The QA lives in a harsh world but that’s the way it is. Repeat faults are anathema to QA and if repeat faults exist it’s a condemnation of the QA system.
In GW2, I would argue that there are multiple examples of repeat faults. Now, I don’t know enough about the internal workings of “the company” to make firm concrete recommendations about their internal QA but it’s apparent that it’s, hmmm, challenged. In order to have a quality product you MUST have a quality QA department. So, with that in mind, I’ll make some generic recommendations that have applicability to QA in general though not necessarily to “the company” in particular.
1) QA structure. QA doesn’t work for the folks they are quality assuring. They are semi-autonomous reporting to the highest level entity with purview and authority to direct changes in the systems being QA’ed. In some organizations, the QA department reports directly to the CEO.
2) QA personnel. QA starts with quality people. They must first be experts in the systems they are assuring. They must secondly be experts in professional quality assurance processes and procedures. They mustn’t be seen by the workforce as “squealers” but they can’t be afraid of calling someone out for shoddy work (and if they do that, they can’t afford to be wrong). Ideally, all QA Professionals would have spent years actually correctly performing the work they are now quality assuring. As a result of this experience and expertise, they get paid a lot of money. Interns make really bad QA professionals (as do summer hires brought in off the street to handle surge requirements).
3) Automated systems (it’s all about transparency). It is essential to have an automated professional system in place for reporting faults, tracking them, cataloging them, suspensing them, and disseminating information about fixes throughout the company. Every individual in the company should have read access to that data. In some situations, I would argue that your customers should have read access as well (though that is probably not appropriate for this environment). Each fault identified and cataloged in the tracking system is tagged with an owner. As the fault is fixed (usually by a process action team) the owner inputs that fix in the tracking system, the QA that first identified the fault verifies it’s been fixed, and then the QA updates the solution in the system of record. Included in that final update is the fix to the process that will prevent recurrence of the fault. That data, from birth to ultimate fix, is then never deleted but is moved to a historical file for metrics and lessons learned. Over time, this historical file provides you leading indicators that you can use to predict faults and prevent them before they occur.
4) Eliminating repeat faults. As a general rule, the person that caused the fault can’t permanently eliminate it. They should be part of the process action team that reviews, assesses, and recommends changes to prevent recurrence. But, the key here is they are part of a team. That team comes from representatives of every part of the process that touched the system from which the fault came. That team determines the root cause of the fault, determines changes to the process to prevent recurrence, and makes those recommendations to management. Once the system is fixed to prevent recurrence then that information is disseminated to ALL affected parties (acknowledgement required).
5) QA coverage. There must be sufficient QA personnel to properly monitor every aspect of the process. If the QA has to cut corners to get something out the door to meet a tight timeline, then you don’t have a QA department. You’re just paying lip service to QA. There are two ways to address this. One, you either extend timelines to ensure QA has time to do their job, or two, you add additional QA personnel and prioritize the work. There’s really no other quality way to do that.
Those are a few, off the top of my head, suggestions. I can provide tailored recommendations if you desire, but I have warn you it would be expensive.