David's point was well-made, but this post makes a good point as well. 37signals has huge structural advantages because of the popularity of Rails and their blog's huge audience. When you can push out your announcements to over 80,000 feed readers alone, you have an advantage few others can match.
The bottom line is that getting distribution is hard. Those who have it often forget that.
Sure they can roll out new apps to a huge audience now, but they did not start with 80k readers. It took them years to transition from a consultancy to a SaaS model, and years to grow Basecamp and Rails. Getting distribution is not that hard, but it takes a LONG time (talking from experience).
people seem to miss this aspect of grabbing success. 37signals took 5 years to get here. wikipedia/wiki/crowdsourcing took 8 years. craigslist took more than 13 years.
success is a balanced equation, you need equal parts marketing, sales, engineering, product design, pricing, timing, and even a bit of luck to get to what you want.
That's true, that they get a lot of publicity for their apps, but that's not an accident: they have worked hard at building up their blog and no doubt paid DHH and others for spending their time on developing open source software. Clearly 37signals realise that marketing is important - and they are good at it.
So I won't try to market my line of small business software.
Instead, I'll just promote the fact that it's so sophisticated, it could only be written in a brand new language that I had to create myself: eJavaLisPython++.
I will soon release a new module for eJavaLisPython++. Its source code will probably be, "jjl43", but I'm running some unit tests to see if I can get measurably better performance out of "jjm4".
It's purpose was purely functional: to make it easier to "meet" the people I've already "met" here. It worked like a charm. Well almost like a charm. I didn't meet you. Perhaps next time.
While this point is true, it is moot to DHH's advice. Why?
Because marketing plays the same roll whether you are trying
to build the next Facebook or simply gain a couple thousand paying customers. It has to be there, and its gonna take time and effort.
As DHH points out in the comments of the post, the final line references a mathematical error and not a fantasy world .
Ultimately, we as internet developers need to start thinking about charging customers. I'm always talking with friends about how applications are moving from the desktop online. And some of those applications best fit some sort of pay for use model. How to get users is one thing, but that wasn't DHH's point.
I think that if you really look at the startup landscape you'll see that there are a bunch of companies trying to charge people (mine included, although we exist in a different domain than most of the startups I run into). The hardest part of that equation isn't the charging, it's getting the buzz out and getting people to even look at your app in the first place.
Just changing the founder mindset to "charge for something" isn't nearly enough.
The idea of charging for web apps has been tried to DEATH. This "revolutionary idea" has come and gone 50 times over. It is extremely hard to make _more_ from this model than through advertising, being strategically aquired, or just about any other model. I personally have tried, hard.
If people aren't in the spending mode when they go to your site then just forget it.
You're forgetting the second point DHH made. Target small businesses. They're a little more likely to spend money to improve performance than your average Joe internet user.
I'm unsure how this is different than any other business, whether online or brick-and-mortar.
He says right in the presentation: this is not guaranteed to work. You still have an uphill battle; the general wisdom is that three out of every four businesses started in the US fail within their first year.
However... your odds of creating a profitable company based on a sound business model are far better than the odds of you building a startup with an insane burn rate and surviving long enough (and creating a service useful enough) to have a liquidity event.
Um, those weren't 37Singals' numbers he was putting up on the screen, you can bet on that. The difference between that and what 37 Signals makes is what can be attributed to marketing, yes.
Saying that they are wholly dependent on marketing to be profitable at all is a little much though.
I'm pretty sure he said getting 500 paying customers is still pretty hard. Just not as hard as getting the millions of customers needed to "be the next Facebook."
The bottom line is that getting distribution is hard. Those who have it often forget that.