Let me ask you something concerning MTBF
Do make compromises around gathering and analyzing data since you only need to report MTBF?
Do you use MTBF (exponential distribution) based test planning when you know the product has a non-constant hazard rate?
These questions came up this week via email looking for advice when directed to ignore the actual situation and just do what the customer wants.
I’m traveling this week, rather jet-lagged today, so going to keep this one short.
How would you answer these questions? What advice would you give someone using exponential based reporting, test planning, or data analysis approaches knowing the customer expects that process yet the data and your experience suggest you should use another method  (Weibull or MCF, for example)?
Please add you comments below and let’s prepare a list of what one should say or use to respond to such actions.
I have to present MTBF, knowing that is complete BS, but I always attach a real test plan, reliability block diagram, R(t) with CL for the first, and second year and expected EOL. So if someboday wants to know the facts, he can find them and ask. I stopped arguing about MTBF, but I can’t stop myself to make a mark like: “Do not make any decisions based on the MTBF value”.
Hi Nils, good for you and keep up the appropriate and hopefully thought provoking comments when you have to document MTBF. cheers, Fred