This really sounds like a bad way to start a software product. You should be choosing an area that you have a real interest in, or at least some strong knowledge in. How are you going to know what your users need if you don't understand the market properly.
Lower customer acquisition costs is absolutely not going to be the reason your software product succeeds or fails, there is so much more to it, don't base your decision on this.
Find something you are interested in and have knowledge in, then see if the market need is already satisfied. If it is not, then you have found your area. Alternatively, if you are awesome, tackle any market knowing that you will execute better, but you better be awesome to pull that off.
Traditionally Military, Medical, Aviation, Banking and Scientific Research... Basically anything that is hard and takes a lot of domain knowledge and huge amount of certification to get into
IE. Anything with a massive barrier to entry.
If you have knowledge in an area like this, then I would say, "go for it" because a good solution here you will be able to charge a lot for.
Outside of that, there are key areas of "speciality" in just about every industry from software development itself through to painting houses, experience that people who do it all day every day have ... you can take, distill and transform into a software solution ... my company has done this in about 25 different industries so far.
Really, if your doing it part time, then you want to pick something either:
You know well and that your interested in... Otherwise you are taking too many guesses about things you don't know OR paying someone who does know the answer.
OR something very general that everybody needs (like 37signals do) but these are very high competition and unless you are a specific user interface god your going to struggle to get the consumer markets attention.
This is an interesting question, which makes a lot of sense, as we all want to maximise our potential for profit. However, you do have to play to your strengths.
Banks would be willing to pay a million dollars for something that helps their trader's P&L, but if you don't have the right knowledge and contacts, you'll never get in the door.
Ok, here we go. Asnwering your question bot totally useless brecause not a nieche one person can exploit on the side.