Wednesday, October 31, 2012


My plan was to depart New York City on Tuesday morning, but Sandy didn't want me to leave quite yet. I spent Monday night safe and sound in Long Island City, but saw the flash of the 14th street power station transformer explosion followed by Lower Manhattan's plunge into darkness.

Looking out of the window, it seems like only the Empire State building has power below 42nd Street. 

Looking out the other side of the building, here's a submerged construction site in Long Island City.

Water had gotten into power lines running underground, touched hot copper elements, and turned into steam.

Saturday, October 27, 2012

The Two Hurdles of App Adoption

To get users onboarded onto your app, you have to overcome two hurdles:
  1. Get people to download the app, which takes annoyingly long.
  2. Get people to sign up inside the app, which they may not complete.

On the web, these hurdles are much lower.
  1. You don't install a website, you just visit it.
  2. Websites can set a cookie to remember you. Facebook's social plugins are beginning to show how to avoid lengthy signup processes by relying on a central source of identity.
I believe that removing or at least lowering these two hurdles on mobile is incredibly important. It will lead to better apps and a level playing field: Right now, the most popular apps are mobile versions of already-popular websites. But these are not the apps that are the best-executed and the most useful. It will also lead to a better mobile web, rather than the one cluttered with "download our app" interstitials.

In the next few posts on this blog, I'll lay out a genuinely new approach to solving app adoption and thus improving the app ecosystem. Stay tuned.

Thursday, October 25, 2012

Rethinking the App Model

The difference between Hacker News and this blog is that tomorrow, you're going to visit Hacker News, but you're not likely to visit this blog.

This distinction does not exist in mobile apps.

Mobile apps are downloaded and installed in an elaborate multi-step process. They're backed up and restored when you get a new phone. No effort is spared to make sure they stick around forever. Each app is treated like a box of childhood photos, a permanent possession.

What you get is endless screens of apps, quickly sampled, never discarded.

Your browser is aware of this distinction. There are websites you visit often - Gmail, Facebook, Hacker News. Maybe you have them bookmarked, but if not they autocomplete in your browser after you hit the first key. Assets and Javascript are cached. Fresh content gets loaded from the server. Sites you visit less often eventually fall out of the asset cache and the autocomplete list.

Mobile apps should move toward the browser model. You hear about something new. A shell of the app is installed and downloaded in one click. Assets are lazy loaded. Each app joins an unforgiving LIFO queue that wipes them out when they're not used for a while. You can pin apps that you want to stick around forever.

The big winners in mobile apps tend to be the companies that first won on the web: Facebook, Twitter, Yelp, Pinterest, and so on. A great website can win because every other site is only one click away.

Websites are easy to get to, and easy to leave. In contrast, mobile apps are hard to get to and hard to leave. They shouldn't be.

Wednesday, October 24, 2012

The Biggest Problem in App Discovery

In my last blog post about the funnel of viral mobile apps, I wondered about a strange phenomenon: "only 90% of of the people who download your app will actually open it - I'm guessing that people are stockpiling apps for the impending apocalypse."

A reader on Hacker News pointed out that maybe this isn't irrational behavior on the part of users, but attention-span slippage due to the long time it takes to load and install an app. This led me to make this little chart. Experiencing new products on the phone - not just iOS as shown here, but also Android - takes an order of magnitude longer than on the web. It's a problem that we should fix.

Tuesday, October 23, 2012

Every Step Costs You 20% of Users

In a consumer mobile app, every step you make a user perform before they get value out of your app will cost you 20% of users.

Here's typical consumer app funnel. It applies to apps that do messaging (like GroupMe or DrawChat), but also posting content (like any app that ends with "-gram" a la Instagram and Cinemagram):

Your exact numbers might vary. With DrawChat, we found that 90% of people that land on the App Store page will install the app. The next step is to actually open the app, but only 90% of of the people who download your app will actually open it - I'm guessing that people are stockpiling apps for the impending apocalypse.

Sign-up is a much bigger hurdle than the previous steps: DrawChat's signup verifies your number via SMS which will get you completion rates around 70%. A startup I've talked to has seen completion rates as low as 50% with forced Facebook logins, while another one allows signing up with different methods (username/password, Facebook, LinkedIn) claims completion rates of 90%. Exact specifics may vary depending on what permissions you ask for.

Depending on the length of your funnel, you end up with 20-30% of the original users actually contributing to your ecosystem by creating and sharing new content.

What to do? There's not really much you can do until users get to the point where they've opened your app. But once a user has signed up, make sure you take them by the hand along the funnel. Make signup simple and unobtrusive. Make content creation easy. Come up with the simplest possible sharing experience. Because of the shape of the graph, you need to get each user to share with 3-5 friends for your viral coefficient to go above 1.


Updates / Reactions

1. mikebo asked on the Hacker News entry for this post how I know the number of App Store page views - it's not something that Apple provides. Excellent question. The answer is a bit involved.

People that don't have DrawChat yet get invited with an SMS that contains a link to a page on our server. That page redirects you to the App Store. So we know how many people see our App Store listing. On the other side, Apple provides you daily download numbers, which you then compare to the "first app open" counts on Flurry, Localytics, or your analytics provider of choice. We can also differentiate people who came to DrawChat organically vs. through invites by seeing if they had an invite waiting after sign up.

2. Shenglong asked on Hacker News if it's not better to start users in an "Explore" type experience instead of asking them to sign up immediately. This can be beneficial, but on the other hand, having an sign-in-less "Explore" type experience may actually hurt your signup conversions. Users get a feel for the app, decide that they'll sign up later, close the app, and never come again. I might be hallucinating but I think Instagram used to have an "Explore" screen with interesting pics pre-signup, and in the latest versions they got rid of that. Does anyone have data on this?

3. udit99 suggests that the 90% download-to-open ratio is not due to people stockpiling apps, but rather "attention-span-slippage caused by slow app downloads." I believe it.

Monday, October 22, 2012

Tablets are for Games and Email

There's an interesting new paper out by some folks over at Google that explores how tablets are used in the field. The top activities by frequency and reach are games and email, followed by social networking, looking up information, watching videos, and shopping. Here's an graph to illustrate this (click to enlarge):

Wednesday, October 17, 2012

Social Vocab 101

I scribbled down this graph with common nouns and verbs in social products:

Almost every social site and app has equivalents of these: Pinterest says "repin" instead of "reshare". Facebook lets you "friend", but also "subscribe".

The basics: Like, Comment, Reshare, and Create. These are roughly in order of occurrence  The dropoff can be up to an order of magnitude with each step: The average Facebook status has far more likes than comments.

People are passive, and creating content is hard. The more creativity it requires from the user, the less likely they are to do it. Write a status? Easy. Snap a photo and apply a filter? Medium. Write an eloquent blog post? Hard.

Great user retention strategies rely on new, useful reasons to notify people. Circa 2007, you used to get an email when someone messaged you on Facebook. In 2012, Instagram notifies you when a Facebook friend of yours has joined. Quora sends you a digest of answers to questions that are relevant to you.

The holy grail: The elusive invite. That glorious moment that pushes the viral coefficient towards 1. Everything has been tried: Gmail brought us the exclusivity of being one of the selected few. Some sites plainly invite you to "invite 5 more friends". Rare is the site like Eventbrite that wouldn't work unless you shared with others. That's the kind of product you want to build.

Tuesday, October 16, 2012

Chart: 40% of iOS Users upgrade to iOS 6 in first week.

Here's a fascinating chart I got in the mail today: The adoption of iOS 6 in its first 11 days.

Two highlights:
  • Within the first week over 40% of users had upgraded. 
  • Almost all upgrades came from iOS 5 users. There is a resistant core of about 11% of iPhone users that remain using iOS 4.

Source: Velti's State of Mobile Advertising Report September 2012

NYC Revisited

I've decided to spend another month in New York City in October and I'm working out of WeWork's offices in SoHo West. Building improvements to DrawChat and working on a couple of ideas I've been developing.

Loving the energy of this town.