In a games studio you do not ask whether a feature is built to spec. You ask whether it is any good, and nobody knows until they have played it. A game goes through many iterations before it works, and decisions are made over a prototype, not a specification.
We worked in three designer and developer pairs, with one testing department. Each pair released a game roughly every two months, so seven came out in seven months. The company was small, which means little structure and a lot of discussion: about features, with the external sound supplier, and above all among ourselves about what good actually means.
I introduced regular demos, a roadmap and a ticketing system. We started writing game design documents and sharing know-how between the pairs. It was not paperwork for its own sake. It was so that three groups of people, each working on something different, would see the same picture and could talk about concrete things rather than impressions.
The work did not end at release. I followed player feedback and how many people were actually playing, gathered opinions inside the company and used that to suggest what to add next.
These were people who lived for their work and brought back inspiration from games events. On a team like that you can see that the gap between finished and good is no small thing.