Hard to say, mainly on account of it depends how far up a creek you are without it. :)
*It maybe PC and all but lotta folk won't wanna talk about failures, in particular it being a Friday and all...
I doubt you're going to get solid numbers but hopefully some fun anecdotes. I'd love to know how often this happens because I've long believed it's a real pain point among consumers that can be easily solved with existing technology.
In today's interconnected world there's no reason recorders/cameras can't send continuous status info to manufacturers or service providers (i.e, integrators providing equpment monitoring and maintenance value add). Essentially, if some recorder or camera in the world stops working, there's no reason we can't know about it and get it fixed before its video becomes of critical importance. So why isn't this as standard a feature in video surveillance as is, say, HD?
Rukmini, this is an interesting topic for general discussion though, as Steve mentions, getting accurate numbers will be difficult.
I guess the numerator is number of incidents / failures to provide video. But what's the denominator? Days? Months? Years? Number of Cameras?
In addition, your AND condition is key - "no recording at all was made AND customer needed footage"
The former condition - no recording available / made happens far far more common than the combination of it being missed AND it being needed.
Overall, I think the pain of this is low BUT there are outliers where it is disastrous (e.g., Murder in Front of Broken Camera Missed).
I would say almost never. I know that is not quantitative so I'll throw out less than 0.1%, possibly even orders of magnitude lower. Seriously, other than the rare total power failures that last so long even our UPS runs out of juice, we are truly running the "five nines".