Althenick and C Mil give good answers I think. For "Build" quality I'd add "design". In very simple terms, there isn't so much trial and error these days due to more advanced design tools.
MoD traditionally keeps this simple and looks at flying rate when predicting attrition. I think you're right about timescale; for example, I remember the Lynx attrition rate dropping from approx 30,000 hours to 40,000 hours in the mid-80s, which meant a planned attrition buy was scrubbed (at around the time of the "Westland crisis").
There are slightly more complex factors, often trends which are resolved. There were a few SHARs lost due to outriggers hitting the new ramps. Absolutely nothing to do with design quality or maintenance. In such cases the attrition would be deemed non-recurring (in the long term) and engineering judgement used to adjust the stats.