Quality is such a fun topic to discuss. Good quality - and bad - is instantly recognizable, but difficult to define. I bet that everybody who has been involved in a translation project, either as a customer or as a vendor, has had discussions about translation quality: how many mistakes there were, what was a mistake and what was just a difference of opinion, what related materials and terminology lists were given beforehand, and so on.
Quality issues can be very frustrating to both customers and vendors, and they definitely affect the profit margins of both. It would be great to have a clear-cut definition of a good translation, a "six-sigma of language". The problem is that quality is very much about perception. Yes, there are the typos and grammatical errors, which are easy to define and to spot. But then there is style, and the question of whether the intended message was conveyed or "lost in translation", like Bill Murray in Tokyo. And that is where the slippery slope begins.
Eventually good quality means that the target audience received the message its provider intended it to receive. In the translation industry, this also includes that the correct message was received in time and within the translation budget.
LISA has done a major effort to establish a QA model which is described on http://www.lisa.org/products/qamodel/. The LISA QA model helps to quantify some of the quality issues, and gives a tool which customers and vendors can share in their discussions. Most importantly, it has been developed in co-operation by end users, software and hardware developers, and localization vendors, so it accumulates their joint experience.
One can, however, quantify only so many qualitative issues. The rest - the more elusive "perceived issues" - tend to fall under customer relationship management.
