As with everything, Jonathan, it depends.
Is the product an internal tool, or public facing? Is it an app, or a content site? Niche tasks requiring expert domain knowledge, or general tasks?
My experience with contextual enquiry has been limited to "lean" studies, where we only observed a handful of users. However, you'd be surprised how much you can learn and what patterns you can identify with input from only 5 or 6 users. If there are distinct user groups, however, you'd want to make sure you had a few representatives from each group. You may want to allow for at least three weeks of elapsed time:
* 1 week planning, preparing, scheduling visits
* 1 week conducting site visits
* 1 week of analysis
In terms of getting people interested, what is it you mean? Do you need to convince management to invest in this research activity, or are you asking whether participants should be rewarded for their time? That all depends too—if the participants are members of the public (e.g. recruited through a screening agency) then you'd usually pay them for their time (e.g. $100 would be normal here in Australia). If it is a co-worker at an organisation, I usually arrange for a couple of movie tickets to thank them for letting me observe.
Hope that helps?