The best way to make consumer internet products

Most popular consumer social internet products sound stupid when they first started. I’d argue that most of them are still stupid. The only difference is that the popular ones gained enough social proof to not look stupid.

So if they’re all stupid, what then separates the products that take off and the ones that don’t? Well, this reminds me of the old ‘hedge funds are all luck’ theory: take 100 hedge funds. They’re run by the same white-shoe, pedigreed types, and they’re all investing according to their own arbitrary ‘investment theses’. 80 of them suck, 10 match the S&P500, 9 beat the market, and 1 looks like Midas. Statistically, someone’s going to luckbox their way to the top. The problem is that you can’t play too many of these at the same time because it’s hard to raise a few hundred million dollars.

Consumer internet startups are like hedge funds. A consumer internet founder pitches that they’ve discovered some under-exploited consumer demand and are delivering a product to monetize it. This is analogue to a hedge fund pitching LPs that they’ve discovered an opportunity to deliver outsized returns and are delivering an investment product to capitalize it.

About 7 years ago, Paul Graham had a good investing idea. He realized that 1.) the cost to fund a startup is getting increasingly cheap and 2.) he only needed a piece of a big one to make his entire fund. In short, it’s possible to bankroll a whole cohort to find the luckbox. Since pedigree and track record don’t matter that much (beyond a certain threshold of competency), it’s more about creating the right basket at the right price. Because the bets are mostly on people who have little track record to demand better prices, we get our money in well. I don’t know how good YC returns are, but they seem like they’re doing well and a lot of people are copying them. Corollary: Don’t invest in consumer internet startups founded by celebrity entrepreneurs. They’re expensive and they don’t have an edge.

So why not take this approach to the next level? This is my strategy: make a mobile app framework with 3 views: a feed page, a profile page, and an “action” page. Notice that this is the same structure as every popular consumer mobile app. An action could be snapping filtered photos to passive location sharing. Corresponding feed page and profile page are collections of said user-generated actions. Poop out a new ‘action’ every day. The ‘pedigree’ of any individual action doesn’t really matter since within a month you’ll have 30 of them. Hedge fund entry tickets start at around 8 figures. YC-style investing lowered the entry fee to 5 figures per bet. My strategy lowers the cost down to 3.

But the story gets even better. Mobile hardware capabilities have been static. We have the same camera, same GPS chip and same WiFi functionality on our circa 2013 phones as our circa 2007 ones. And that remains so for the foreseeable future. Thus, the problem space is finite and static. This problem seems tractable, so why not algorithmically generate the different permutations of ‘actions’ in the space.

Easy. Let’s start minting Instasnaps.


Now read this

A new adventure

Today is my last day at Groupon. It’s been fun these past 13 months – with all the highs and lows that come with selling a startup to and working for a company re-discovering its legs after falling out of favor with commentators that... Continue →