Sam Gerstenzang quotes Snap’s S-1 filing and adds his comment:

Our strategy is to invest in product innovation and take risks to improve our camera platform.

I can’t state strongly enough how differentiated this is. No other company who has reached this scale works this way. Their long term differentiation is not based on a classic ‘moat’ but instead on the velocity of product innovation. Snap operates like a toddler running head-first down the gangway. Without momentum, Snap believes it will fall flat to the ground. In other words, Snap doesn’t believe network effects will save them. In our current App Store mobile world, where you can launch a new social app by authenticating it against the address book, network effects are reliant on the value of aggregated historic content, of which Snapchat has very little.

Another extract struck me:

Early on, we thought about charging our community to use additional features. For example, we saw so many people having fun with the Creative Tools we made, like drawing and captions, and we thought people might want to purchase additional ways to express themselves. To test this hypothesis, we built a Lens Store where our users could buy new Lenses, in addition to the free ones we already provided. The results were disappointing. Only a small number of people wanted to buy Lenses, and the number of people using Lenses decreased. After a few weeks, we got rid of the Lens store and made all of the Lenses available for free. Almost immediately, our community began to use Lenses more and create more Snaps to send to their friends and add to their Story.

This short description of Snap’s process is actually extraordinarily revealing. Snap is so good at product innovation precisely because they are okay with being wrong. They try large risky things, take feedback, and try again. They’re perfectly ok with launching and then rolling back major product changes (like Auto Advance) because it taught Snap (and their users) a new way of interacting with the product, which they eventually figure out to execute well (in the form of Story Playlist.)

Small experiments like these are not a new idea. Good reminder though.