COMMENTS
Dharmesh, I absolutely agree with your views on infrastructure, but this company did not go out of business simply because if that - that's pure BS. My argument became too long for a comment, I posted it on my own blog:
http://www.zoliblog.com/blog/_archives/2006/6/30/2077022.html
Dharmesh, I really liked the "pricing discontinuities" section of your previous article, where you talked about each of the points in time where you had to jump to a more expensive hosting environment. Would you mind going into more detail about when you knew you were ready to make each jump, and the cost associated with each? (I guess I'm kinda wondering, what exactly are you paying $2,000 for right now? And more generically, when should I forecast moving up each level in my early-stage business plan?)
I think your point #1 misses a crucial part of the "Web 2.0 attention economy": advertising. If you run some effective ads, eyeballs do in fact pay the hosting bills.
Ryan - what "previous article" are you referring to? I would like to know about the $2,000 hosting fee too to help me plan. Thanks.
Stacy, I'm thinking of "Startup Reality Distortion #5: Web Hosting Is Not Free" from May 26, which is available at http://onstartups.com/Home/tabid/3339/articleType/ArticleView/articleId/620/Default.aspx
You're getting ahead of yourself. HubSpot doesn't need a $2,000/month host, it needs a $10/month host. A closed beta doesn't need a host like that, you're simply throwing money away. I agree that to an ESTABLISHED business with a customer base a $2000/month host is very important but not a startup without paying customers. That money is better spent improving your product and getting the word out. The time will come when you need a dedicated server, it's just not right now.
Nolan: Though it may not seem that way from the current website, HubSpot does indeed have paying customers. These customers already justify the hosting expense.
Initial customers take more hand-holding than I would like (hence the delays in a wider beta). But, that's a topic for another day.
OK I believe you have paying customers and that's great but again I go back to the heart of my point, startups don't need $2,000/month hosting, they need the minimum. CouchSurfing was online for awhile and did need a better host and a better plan. I agree with most of your posts but your views on startups and money sometimes baffles me. Get the product, then the customers then spend the money. It's easy to spend money before you need to, if you don't spend that $2,000 when you don't need to that is $2,000 more to your product or even better, $2,000 less debt.
Nolan - that's easy to say but it's not reality. You can't get "paying" customers without a working site hosted somewhere. The type of hosting you NEED depends upon the web application. Simple sites can use cheap shared hosting, I agee. But if your app requires things like a windows service running, or programmically creating folders and such, shared hosting won't work. So without knowing specifically the needs of the app, saying that $2000 is out of wack - is out of wack, my friend.
You're right Stacy. It does depend on the needs of the application and highly complex applications need appropriate hardware, bandwidth and storage as do small applications. I said startups need the minimum, in some cases that will be a $2,000/month host, in most other cases it will be cheap, yet well managed shared hosting until they need more.
I am trying to advise some of the less technically proficient readers that don't quiet understand the subtleties of what they do and don't need as a startup. You wouldn't believe how many businesses I have worked with whose first order of business is buying a server and spending extravigant amounts of money on hosting before a single line of code is written.
I'm the last one to advocate spending money too early on things like web hosting.
HubSpot is a special case (we've been writing code for over a year, have paying customers and adding new ones every month).
The application itself is also relatively robust (it does more than just track bookmarks). So, it takes a bit of horsepower to make the magic happen.
But, I'll repeat: If you're a startup and have not "launched" a product yet, you should be able to get by with < $200/month in hosting fees.