Gray and Rob make great points about thinking through why you want a patent before pursuing one. If it's just going to hang on your wall, you can get some nice art that costs a lot less money and looks a lot nicer. :-)
One thing to keep in mind re: figuring out the business first, though, is that a lot of those early "lean startup" activities (getting out and selling, talking to potential users and customers, etc.) can trigger legal bars that preclude later patenting. And oftentimes it can be the early innovations that end up being core aspects of the business that one would later like to protect.
Fortunately, while the patent process CAN BE very expensive, there are startup-friendly solutions. You can apply the "Lean Startup" methodology to the IP process and use things like MVPs (Minimum Viable Patents), liberal use of provisional applications (potentially with the entrepreneur using attorney-provided templates and guidelines to take the laboring oar and minimize spend), omnibus filings for multiple inventions, the Patent Cooperation Treaty internationally, etc. All can help preserve optionality to build a patent portfolio with business value, while deferring the bulk of the costs until you actually get the business figured out.