The Manual-First Startup

Agree 100%. As an example, when I first created Cocoa Controls, I wrote a little Ruby app that iterated over every single one of GitHub's Objective-C repositories and stuffed them into a database, where I then manually checked each one to see if it was worth including. Literally every repository. By hand. It was a huge amount of effort, but it was also the only way to get the amount of content that I needed to make the site useful enough to get contributors to submit content.<p>Nowadays, the content volume is more than self-sustaining (I have 46 different repositories in my queue, all submitted by readers), but I still check every submission, clean them up by hand, and manually publish.<p>(<a href="https://www.cocoacontrols.com" rel="nofollow">https://www.cocoacontrols.com</a>)

At first I thought this was going to be like "write the manual first" a la Fred Brooks. But it actually made a great point. Yesterday for the first time I read pg's outstanding growth essay <a href="http://www.paulgraham.com/growth.html" rel="nofollow">http://www.paulgraham.com/growth.html</a> which confirms many of these same excellent ideas.

Excellent advice and reminds me of the Human Intervention as a Competitive Advantage article from Derek Sivers: <a href="http://sivers.org/hi" rel="nofollow">http://sivers.org/hi</a>

Related: <a href="http://ryanhoover.me/post/43986871442/email-first-startups" rel="nofollow">http://ryanhoover.me/post/43986871442/email-first-startups</a>

tl;dr Do non-scalable things to reach product/market fit, then do scalable things

I'm curious. My understanding of anything-first startups is that you do everything manually to test a market, grow your clientele, and turn a profit. In your post, you mention not being profitable yet, even with the injection of around 7.5 million dollars. As someone who is now used to building businesses that turn a profit <i>before</i> they launch, your situation is interesting. Note that I'm not looking to start a flame war, just genuinely interested.

Great point. This is, in business, is much the same principle as avoiding premature optimization, in coding.

Definitely agree on the virtue of doing stuff as manually as possible. Really hard though mentally as you always want to do things the right way. I guess at such an early stage though, the right way is whatever gets you feedback quickest.

Appreciate the post; have my own complex task also involving sorting that I was tempted to spin my wheels on building a tech to accomplish. Think I'll forego, and instead collect data on manual sorting instead.

Very good blog post, if there was a startup manual, this should be in it. :)

That was awesome and showed a high level of grit and audacity. Kick. Ass.

Shameless offtopic plug: avoiding having to write crawlers and scrapers is exactly the problem we at <a href="http://import.io" rel="nofollow">http://import.io</a> hope to solve.

See also Derek Sivers, Version 0.1 = Start lo-fi <a href="http://sivers.org/lofi" rel="nofollow">http://sivers.org/lofi</a>

here's another one: <a href="https://teelaunch.com" rel="nofollow">https://teelaunch.com</a>