John brings up an excellent point - you can't simultaneously fix scope, time, cost, and quality, and it's usually on scope that you'll want to bend. Discipline yourself to prioritize the features which really need to go into your product NOW, and put the rest aside for the next release (this is easiest to do in software, but applies elsewhere as well). Don't be afraid to drop things from the pipeline altogether if the market indicates a better direction for your product, even if you've already put a lot of work into them. Remember that you're in a continual process of finding out exactly what your customers will buy.
I don't think being beat to the punch is as significant a threat as most people make it out to be - unless you've given competitors enough time to become *really* entrenched, you can still beat an incumbent with a better product or user experience. Releasing in one month or two is unlikely to make the difference. More troublesome are systematic changes to the landscape: let's say you announce a Google Wave plugin right before it gets discontinued. Whoops, better pivot quickly.
Most failures I've observed have multiple contributing factors, but one of the most recurring I've seen has been a lack of discipline on the founding team: jumping to a new product area too quickly - or too slowly. Not filtering ideas enough, or filtering out too many. One part of the team having unrealistic expectations of the rest of the team (you can only sell as fast as your company can service those sales; asynchronies in your pipeline can put a lot of strain on your company and team if you try to follow through on them). Sometimes simple lack of compatibility.
I've experienced some hurdles personally as well, usually in team composition: going it alone when I really needed a partner, or going in with partners when I needed to go it alone (it's a misconception that you should always have multiple co-founders. There are some businesses that you really should run on your own; my first success was one where I made the judgment call to deliberately run solo). In a case where I did need a partner, getting so desperate that I selected the wrong partner - that's really hard to recover from.
Going into a two-sided market. Think long and hard about how you're going to get one group on without the other before you engage in any sort of matchmaking.
Going into a heavily regulated market: investors don't like regulatory risk. Don't do this until you have a lot of cash upfront, or you'll encounter a circular dependency between funding the regulatory filings and getting regulatory clearance so you can raise funds.
Just keep learning, when things go wrong - and when they go right.