As a general rule, one chooses for his subject the broadest and most general version that he can feel comfortable with. This almost always involves broadening the scope of what one really wants to say. The problem is that one naturally divides the subject into what is known and what is not known, and works at the interface. To your audience, however, almost everything is unknown: they have no feeling for where this interface is, or how the problems at the interface arise from the larger context. Thus, instead of talking about killing tensors, one could talk about conservation laws for particles in general relativity (including perhaps, some discussion of spinning particles, charged particles, what happens when a particle breaks into two particles, how this subject is related to conservation of stress-energy, etc.). Of course, what is the most general version one can feel comfortable with depends on how specialized one's audience is. (Thus, for an audience of non-relativists, perhaps 90% of the discussion will be on issues one does not actively think about on a day-to-day basis; for an audience of relativists, perhaps 60%; and for an audience of specialists in one's own sub-field, perhaps 35%.).
The next step is to think of a title. Most of your audience will probably decide whether or not to come based solely on this title. Ideally, one wants a title which indicates what the subject is, what the level of the discussion will be, and which is lively and friendly without being cute. Questions and assertions often make good titles. Of course, one should use no word in the title with which one does not expect one's audience to be familiar. Thus, for an audience of relativists, ``Linearized Fields in a Kerr Background Metric'' sounds technical, ``Perturbations of the Kerr Solution'' sounds dull, and ``Black Holes are Stable'' sounds good.