[bakztfuture]

I don't want to get too much into the weeds, but there is a whole subset in Information Retrieval which relates to IR system evaluation, or search engine result evaluation. One simple way of doing it is simply labeling the accuracy of each result via human curator as either a 0 or a 1.

But it can get really complicated, for example, sometimes there just aren't relevant documents in the index in the first place ... so, you can't really blame your ranking factors too much. The opposite can happen too, where a word occurs too frequently in which case you might resort to other kinds of ranking factors (most notably pagerank).

In our case, we're focused on broadening our state/country level coverage right now for keywords (more listings), then we're going to focus on making sure our location accuracy is a lot better (it needs work). Overtime, you should get the results you're expecting more often :)