Apparently, up to 62 percent of IT projects fail. Some of the reasons why also apply to start up companies and even web startups as well. The key it seems is ensuring you have the right priorities. Not that anyone is implying that this is easy... but what should those priorities be? This article condenses a lot of insights from very smart people into 3 points.

1) Who is your customer?

What is it you are trying to solve and for who? It is interesting to look at successes and failures in the business world and open source world and examine whether they had this right or not. Google was very focussed on providing a better search engine based on better computer science. Honda is notorious for focussing on building very good engines and outsourcing everything else. Walmart was very focused on efficiency in operations.In the open source world, it's clear Ubuntu/Cannonical are all about ease of use. MySQL was about database for web apps and small(er) systems.

On the flip side of this, from my own personal experience, I can recall a distinct sense of fuzziness from my final few years at Nortel. Were we making products that were build like tanks (homage to rich DMS heritage)? Were we trying to drive down costs and leverage off the shelf hardware, open standards, and open source software? Were we going to be a leader in voice, wireless, data, optical, enterprise? The answer was a fuzzy shade of in the middle for all the above. If the market is doing well, you can get away with this for a while. If not, even though you have many very talented people (as was the case at Nortel) then your days are numbered.

There are many many examples here. Feel free to share better ones via. comments.

If you know what it is you're trying to solve and for whom, it makes it a lot more clear how to get in contact with them. This helps beat the fact that No one cares about your stupid little startup.

If you understand this, your longer term strategy will take care of itself with a little analytical thought and examination of trend data. But you must not lose site of:

2) What really matters most *today*?

As Thomas' article points out: work from large to small, scale later, build half a product, essentials only, etc. His point is that in the early days, time and money are more than precious. If you work on that which matters most, you've got the best chance of being around to fight another day and tackle the items that are next most important. If you spend too much time, attention, and energy fiddling over unimportant tasks like fonts and just-so alignments of widgets you're cooked.

For stuff that doesn't matter, see if someone's already done the work and it's good enough to reuse it. This is where open source provides a lot of value. 

If you've made it this far and your plan is still solid, next:

3) Figure out what makes you special

OK, so you're going to make money doing whatever. What is stopping someone else from doing the same thing the very next day? Why isn't someone doing this already?

What special knowledge do you have? What special access to customers/partners/distributors? What patent gives you an exclusive monopoly for a while? What top secret business model innovation do you have? What is unique about your location that gives you advantageous access to your raw materials and factors of production? What uniquely talented employees do you have? Do you have the passion to work four times harder than anyone else for as long as it takes? 

Once you come up with what it is, it's a very good idea to bounce it off someone who will tell you straight up if you've been smoking the good stuff or not.

If you're struggling to come up with something concrete here... or if your reason is because you're smarter than everyone else, save your time and money and go do something else.