It's good enough for a lot of big companies to offer APIs for the enterprise... So yea, it's a great business. What those APIs do and how you can sell them is an altogether different story.
For what it's worth, I started off looking at my technology as an API. What I've built is quite complex and the technology works. I got really good feedback and traction. Lots of people are still very interested. However, I realized I needed more than just an API. Two reasons:
#1. Most important of all - when a client needs to integrate an API, it's not necessarily "easy" ... They need to have developers on hand and need to actually do work. This takes time. This takes money. It's asking a lot and due to the cost for the client you may lose out.
#2. More of an observation than a reason, but interestingly enough more people responded more excitedly (I never had negative feedback by the way, but more or less excitement - sure) when I showed them a "dashboard" ... Everything has to be a dashboard. Dashboard, dashboard, dashboard. I'm tired of hearing the word. =) However, people jump at that and they jump at having 50 million different dashboards because everyone will then tell you that you need to focus and do one tiny thing really well - even IF your services cover a lot more and your API is a lot more robust. They're still going to look for that very simple dashboard and they believe people are going to be ok buying 5 slightly different dashboards instead of one that does 5 very related things. Really interesting thing that I discovered. For what it's worth.
I still believe that APIs can be very powerful because sometimes the data they provide can be used for many different businesses. So if you find yourself in a situation where you're having trouble with building out a very pointed service, then yea...I don't think it's a bad idea. PROVIDED you have strong positive feedback for the features. I'd also go so far as to also say provided you have a few clients lined up who aren't afraid of getting their hands dirty.