In today's world, the largest risk in a consumer internet startup is typically consumer adoption. VCs ask ourselves the question - how many users will the service attract, engage, and monetize and how quickly and easily can this occur?
One simple design factor I've noticed that can be influenced more than others is the position where the product requires user investment (in terms of time) to receive benefit. The simple premise is that human nature is such that we are more willing to engage in products that require little upfront investment. Even if the short term cost will produce 5x returns in future use, we are moving too fast around the net to make this tradeoff - we like to find the path of least resistance. That said, we will tolerate increasing time investment as we use and get addicted to the product.
Too many products require a significant investment of time upfront (e.g. long registration forms, entry of preferences, creation of a large profile, etc) before any value is received. The consumer may be excited about the proposition (e.g. meet new friends, save time doing X with our service, find the lowest prices) but attention spans are increasingly tuned to short term benefits (e.g. search results from one query vs a multi-faceted advance search). The "UI" was even worse in consumer products which required reading a 25 page manual before you can benefit - of course, now the smartest product manufacturers include a one page graphical quick start guide which I would bet is used far more often than any other manual.
One way to solve this problem is to collect only what's absolutely needed initially and then gradually ask for more information - even if the product won't be as flexible or sacrifices some of the long term benefits. These changes seem easy but have huge implications on databases, customer tracking, etc. For instance, at Snapfish, we first made the user do a lengthy registration process before uploading a photo to snapfish. After we noticed the dropoff, we thought it would be easy to allow upload first and registration later once we have the user's photos and they take an action which requires registration (e.g. buy or share photos). It seemed simple but it required us to rethink the way we store customer data as some users would have "orphan" accounts and this issue rippled thru all processes that involved customer data - needless to say it was very tough to even test this change. With new services, its a lot easier to "do it right the first time" (we called it the DIRFT process) - which is advice I give all our startups.
In summary, think about the fastest path to user benefit with least cost/pain (time or data investment) - even if it sacrifices functionality or long term benefit. Once you get a user hooked, you will have their attention to move them up the cost/benefit continuum.