From the worst things startups do at Scobeleizer I think he saved the best points for last in his list (below) but click the link to read the whole thing, fantastic post:
"10. You say yes too often, particularly in engineering decisions. Look at Posterous (the video embedded on this blog). They have a blog publishing tool. But are their comments threaded? No. Will they be eventually? Yes! Why didn’t they do them threaded up front? Because they set priorities on other things that mattered more. That’s actually a good sign for a startup: if you have only four engineers you can’t do everything. If there’s one thing I like about Evan Williams, founder of Blogger and Twitter, is that he doesn’t try to do it all. In fact he prides himself on NOT doing things. It takes great leadership to say “no, Scoble, you can’t have more than 500 members on a list.”
11. Startups pick old technology because it’s familiar. You’re a startup, you should be picking the best of breed for everything you do. If you are using Microsoft Office “just because” then you are making a mistake. Have you considered Jive, SocialText, Zoho, Google Docs and Spreadsheets and Wave before making your choices? Have you really looked at ways to make your small company more productive? Or you just going with the same stuff your dad’s company used?
12. You don’t change direction fast enough. Every startup should be looking at its direction every month or so. Are things going according to plan? If not, fix them. But sometimes you just made bad assumptions about what the market would want from you. That’s OK! But don’t take a year to change directions, change quickly and you’ll have a chance to save the company."