Malware Prevalence

This is primarily intended as a question for you, the reader, to weigh in on:

In this era of dozens of new variants of common malware families, what is the most important aspect of prevalence to you? Prevalence of individual variants, of specific families, of meta-families?

Historically, prevalence has been based on individual variants, e.g. X number of samples of W97M/Melissa.a or of W32/Bagle.g, etc. As these examples show, this also was primarily based on viruses as it indicated the virus' own spread, rather than counting how many ways a trojan author could spam his own creation.

As frequently-updated trojans and bots become more popular and mass-mailers become less so, it may be that this model needs to be revised.

Do you consider things to be more dangerous or notable when their particular attack-vector is being widely used
("Beware – we see an increase in viruses using a specific vulnerability!"),

or when certain types of malware become more common
("Downloaders are increasing in popularity!"),

or is it still most valuable to you to be alerted when just one specific variant is making the rounds?
("W32/MyLunch.m@MM is everywhere!")

Why do you find that to be the most important aspect? How will that information best assist you in protecting your environment?