This week I’ll be on stage at Mobile Future Forward to talk about the “New Waves of Innovation” given the Connected Intelligence era we are entering. While I’m excited about engaging in this important conversation, I wanted to share an observation I’m having as I prepare for the conference. Control over software distribution has never been so tightly restricted in the history of commercial computing. This is deeply troubling for a number of reasons, but I am hopeful for a change in this trend.

If we take a step back to the early 1980s, we can recall a wave of commercial computing where Microsoft possessed an incredible position of power over software distribution. However, Microsoft did not have unilateral control given the architecture of the time – client applications with no network connectivity. This meant that physical retail, direct sales, and bundling partnerships were all viable healthy distribution channels for software providers. In fact, the physical retail channel was incredibly effective and thus a competitive merchandizing driven channel that is an interesting footnote in the context of today’s App Stores.

Jumping into the 1990s, software distribution was forever changed with the emergence of network connectivity. With the benefit of hindsight, vertically integrated networks, such as Prodigy and AOL, were short blips that did not have much impact on software distribution (though again, another interesting footnote in the context of App Stores.) Instead, the web’s open connected nature shifted software distribution online for consumer software. Netscape, WinAmp and Skype are good examples of upstart software products that benefited from online distribution without centralized control. It is worth mentioning that businesses still primarily bought software via direct sales or retail channels. Yet, I think that can be attributed to pricing (fine-grained control) & file sizes (download times given bandwidth constraints.)

In the 2000s, software distribution changed with the move to Cloud-based applications that do not have downloaded components. When this occurred, the catalog of software, and software-enabled content (blogs et al) exploded beyond human comprehension. I would say that this was by far the biggest inflection point of software distribution as the challenge shifted from getting bits to customers, to getting customers to be aware of the bits. Thus online advertising, viral marketing, and inside sales superseded the historical distribution channels of the 1980s and 1990s. As a result, new types of control emerged in the form of black box algorithms that were under unilateral authority of a single entity. For the first time, someone could effectively “banish” an application from the world by hiding its existence. Luckily no one player had a perfect monopoly on online attention, so this was more of a theoretical risk (except when not).

More recently, software distribution has been defined by App Stores that are vertically integrated into technology platforms, and so closed ecosystems have emerged. For the first time ever a single entity had the ability to banish an app as well as a perfect monopoly over distribution. We’ve entered a world where Apple, Google, and Facebook have perfect, or near perfect monopolies of closed ecosystems, each with over a billion users. While these ecosystems contain different distribution characteristics brought forward from past eras it is obscene as to how much unilateral control their owners can exert. What is really troubling is that there are no checks and balances in place. Especially given that these closed ecosystems are only growing in size, as four billion people get online for the first time in the next few years.

As I’ve internalized these observations, I’ve shifted my focus to see what could change the control dynamics structurally. As a VC, I see that the most obvious approach would be to invest in a challenger to one of these players, but that doesn’t change the dynamics. It merely changes who is in control (for example, Xiaomi.) I am wary of attempts to commoditize similarly situated ecosystems since seeing it fail so spectacularly first hand with Facebook’s HTML5 strategy though I will watch Microsoft’s new attempts. And of course there might be regulation put in place by various governments.

However, as a technologist I’ve realized there is one very interesting approach available – to leverage Android’s inherent technical openness to remove central control. Fascinatingly, this has already happened in China where Google does not have control of the Chinese Android Ecosystem. Regardless of the mechanism, or how it is triggered, I am hopeful we are approaching a point of change. I’d love to find startups that are thinking about facilitating the change through their business strategy or envision the change as an unfair advantage to building a big company.

Part 2 of 2: Should you build for the Apple Watch platform?

It was no coincidence that I shared the framework I use to evaluate new platforms at the same time as the Apple Watch was announced last month. My intent was to set the wheels in motion and ask the question that product teams, CEOs and VCs are now facing on the eve of its availability: What version of the Apple Watch should I get? Should we prioritize building an Apple Watch app?

My informed answer is yes.

How did I get informed? First was through holding the Apple Watch and experimenting with the UX in a private demo. I was further influenced by discussing a friend’s perspective after he wore the Watch daily for a few weeks, and hearing how it changed his behavior in a short time. Apple is going to sell a lot of watches and people are going to enjoy using them.

Let’s revisit my platform analysis framework question by question to show why I think prioritizing the Apple Watch platform is a good bet:


1. Technology enablement – Can something be done that wasn’t possible or easy to do before? 

The Apple Watch enables app developers to do many things that have not been possible before such as:

    • New user interface design surface area via Glances, Apps & Notifications
    • New user input devices via Force Touch and Digital Crown
    • New user feedback methods via the Taptic Engine
    • New fitness related sensors and data such as a user’s heart rate
    • New communication methods such as sketch, tap, and 3D emoji

Of course the unanswerable question is what “job needs to be done” with these new capabilities.  The most obvious is rethinking notification experiences for current iOS applications from the ground up. But here’s the truth: we can’t predict the most valuable use cases. We can only predict, as was the case with the iPhone and iPad, that developers will invent great stuff like Uber, Whatsapp, and Roll. It will be exciting to see how developers use the Apple Watch technology building blocks to craft new experiences that weren’t possible before.

 16585696370_8b9e0e2905_k2. Distribution – How does the platform help you gain new users and engage existing users?

The Apple Watch is garnering the strongest cross promotion of any new hardware product line that I have seen.  Further compounding Apple’s own cross promotion of the Watch is the collective media, pundit, and technology interest that will result in even more exposure for early apps.  Thus, breakout apps on the Apple Watch should benefit from the “rich getting richer” flywheel for this early time of the Apple Watch’s lifespan. Based on the high number of people pre-ordering the Apple Watch, this new channel will be constrained only by supply and not demand.

Further, it is worth understanding how the Apple Watch captures a user’s daily attention is tied to the notion of distribution. This is a valuable asset given it is one of the few “fixed resources” in a digital world.  There will likely be a vigorous debate about how much and how fast the Apple Watch can capture attention.  However, my expectation is that it will be a healthy amount within the first year of launch in aggregate and on a per Watch basis.  Thus the developers that can grab as much of this new attention surface area will create meaningful enterprise value.

A somewhat crass but perhaps more apt assessment of the tactical distribution value is that Apple Watch has a new advertising unit for any iOS developer and it is called WatchKit. The cost per install or registered user is unclear, however, the cost per impression is going to be the cheapest on iOS bar none.

2566992664_977268ded8_b3. Business model – Does the platform provider have a clear business model that you can align with to sustainably share in the value created?

This one is easy.  Apple makes lots of money by selling hardware at ridiculously good margins.  Developers of software apps for mainstream consumer scenarios are incredibly well aligned with their business model because, as a platform provider, Apple needs developers to build killer apps to help sell their hardware.

Any developer looking for an outsized growth opportunity should be prioritizing an Apple Watch app right now.  It’s a rare and unique opportunity where success will be determined by those building early great apps in a valuable vertical.  Of course, capitalism and market dynamics mean that few developers who will get this right early will capture the lion’s share of value.

I can’t wait to see what gets built.

Twin Prime

Today I’m thrilled to announce my fourth investment, Twin Prime, is launching.  I vividly remember first hearing about Twin Prime from my friend Semil Shah. Shortly thereafter I asked one of Twin Prime’s early investors, Rohit Sharma, out to coffee to learn more.  Over coffee with Rohit, I found out that one of my old friends, Om Malik, led Twin Prime’s seed round. Here I was, already very excited about Twin Prime before I even persuaded founders Kartik Chandrayana and Satish Raghunath to meet with me.

As the founders explained their insights on last mile mobile acceleration I knew I wanted to be part of their extended team. I could tell they both possess the key characteristics DFJ looks for in entrepreneurs. First and foremost, Kartik and Satish are those special types of founders that combine domain expertise (both have PhD’s in Computer Networking from RPI), deep professional networks (they worked at Cisco and Juniper), and passion to build an iconic company (bringing great company builders around them such as Anand Rajaraman and Venky Harinarayan at Miliwave Labs as well as Rajiv Khemani at Moment Ventures in addition to Om & Rohit). Secondly, and just as important, they had a brilliant insight around mobile data acceleration that improves the mobile experience for consumers, businesses and network owners. I intuitively knew everyone who had a mobile device would eventually benefit from Kartik and Satish’s approach.  

Okay, okay I also really liked the name Twin Prime. 🙂

As a former Product Manager for Mobile at Facebook, I experienced what happens to product and business metrics when performance lags – they go the wrong way!  More viscerally, as an end user, I am frustrated by intermittently slow mobile apps in a way I’d never felt on my laptop.  As I reflected on the fact that a vast majority of Internet consumption was shifting to mobile devices with wireless data connections, I knew there was a huge opportunity for Twin Prime.  I’ll spare everyone the geeky details and focus on the most important thing to know: 

Twin Prime makes mobile apps fast.  Everywhere.  

If you have a mobile app try Twin Prime and thank me later. 

Part 1 of 2: Should you build for the Apple Watch platform?


In anticipation of the Apple Watch launch, many are considering whether to make a bet on the new platform. Will it be another Apple home run product with significant new platform opportunity, or not? This is part one of a two part series with my take on that question.

Over the years I’ve been using a framework to evaluate new platforms based on my observations from working at Facebook and Microsoft, two companies responsible for creating, managing, and at times, mismanaging great platforms. The framework involves three factors and associated questions:

  1. Technology enablement – Can something be done that wasn’t possible or easy to do before?
  2. Distribution – How does the platform help you gain new users and engage existing users?
  3. Business model – Does the platform provider have a clear business model that you can align with to sustainably share in the value created?


 Technology Enablement

Technology enablement has been the historic core focus of companies building software platforms. Microsoft’s DOS and Windows are two examples of platform design driven by technologists on both the build and the buy sides. Historically, APIs, tools, documentation and compatibility were the focus of the platform owner because that is what technologists cared about. As computing became more mainstream, technology enablement shifted to table stakes for creating a valuable platform. This was further accelerated when open source platforms such as Linux, MySQL, and others extracted no value for using their technology. Adding GPS to smartphones, to give you my favorite example of technology enablement, has spawned multiple new killer applications, like Waze and Uber, that weren’t previously possible.



For technologists this was often the least obvious factor to evaluate, but as any entrepreneur will tell you, “they will NOT come if you build it”. Over time as distribution (nee Marketing) has primarily moved to digital channels, ownership of customers has shifted to the platform owners instead of traditional intermediaries like retailers, wholesalers, etc. Prior to this disintermediation, platform owners offered co-marketing, referrals, default placement and financial incentives as the key forms of distribution given the complex value chain. After this disintermediation, Facebook and Apple took these concepts to new levels of scale with Newsfeed, Notifications, and the App Store, as central features within their platforms. Generalizing beyond these examples infers that platforms with direct customer relationships can offer structural distribution, such as access to friend lists and address books, or they can offer surrounding distribution such as merchandising in an app store.


Business Model

One of the biggest risks for app developers is that the underlying platform will subsume their app’s functionality into the core platform. Newer platforms, such as LinkedIn and Twitter, introduced another version of this risk with their shifting policies around acceptable platform use cases which can largely be attributed to misalignment of business models (e.g., viral messaging spam causing user churn or building a competitive product that uses the platform’s proprietary data). Apple has shown the wisdom of designing a business model into the architecture of a new platform. By mandating use of the in-app payment solution with a fixed revenue share and clearly defining use cases that were off limits they created a sustainable app economy worth tens of billions of dollars. If a platform owner isn’t thoughtful about aligning its business model with developers, it’s possible for tremendous value to be quickly created and destroyed as developers abandon platforms quickly.

I’m extremely eager to learn more about how Apple has designed the Watch platform for application developers. However, until the March 9th event occurs, betting on the Apple Watch is a speculative investment for most. Given Apple’s overall iOS ecosystem across iPhones, iPads and soon the Watch, it makes sense for current iOS application developers to take the risk of investing in the Apple Watch platform because of the overall iOS distribution incentives Apple controls. Once I have a chance to understand what is announced I’ll likely follow up with an assessment using the framework I outlined above.

I’d like to thank Paul Davison, Roy Bahat & John Battelle for reading an early draft of this post.

The Three Skills of a Great PM

TL;DR: Good and Great PMs differentiate themselves by focusing on improving the rate of various aspects in product development. Good PMs improve rate of execution and rate experimentation. Great PMs also improve the rate of idea validation. 

Recently a few founders I work with wanted to discuss what a product person does and how to tell a good one from a great one. After thinking through it I decided to share what I’d come to believe makes a good vs. great PM after time at Facebook, Microsoft and my own (mostly failed) startups. More importantly, I’m very curious to hear how others identify great PMs.

5142145393_48a8baa876_bSkill One: Rate of Execution

Fundamentally, a PM helps a company more quickly define, build and launch new products. Entry level PMs start off by writing specs, triaging bugs / issues and helping project manage, with engineering management, specific releases. More experienced PMs take responsibility for successfully launching new products which requires coordination and decision making across a wide constituency such as executives, marketing, legal, finance, etc all while never letting the product’s soul be compromised. While every company has its own specific approach to product development a skilled PMs ensures that every successive release happens faster with better results.

This is really hard and the number of PMs that can do this is surprising small given the number of people with PM titles. Why is it hard? Because often the key to becoming faster involves changing the way things are done in areas a PM has no authority over. Specifically, sometimes it can mean changing composition of the engineering team and others times it means persuading various stakeholders that the plan is the right one so work doesn’t have to be redone. It can even mean getting the CEO to change his mind or even just flat out telling the CEO “no”.

Not surprisingly it is often not until a PM has had 5-10 years of experience building & shipping products that they can reliably accomplish this level of performance. In a startup this is the minimum level of competence that should be hired as an individual contributor PM. Needless to say when evaluating PMs measuring their impact on rate of execution of their team is critical.

4393914288_68c10a576c_oSkill Two: Rate of Experimentation

With software product delivery happening continuously via the web it has become critical for the product development process to include rapid experimentation to validate assumptions & features before committing resources necessary to launch them. PMs start building this skill by learning the mechanics of experimentation that includes multi-variant testing and how to construct good experiments. Experienced PMs shift their focus to increasing the throughput of experiments their company can run as well as the number of experiments that can be run in parallel.

Throughput is surprisingly hard to affect and is even harder to be disciplined about since it is much more fun (and easy) to argue about which experiment to run next. Naturally, everyone thinks their own idea for an experiment is high priority when in reality it is more important to just get the company to run more experiments faster. To really affect the rate of experimentation good PMs have to focus on technical infrastructure that enables experiments, cultural norms around experiments and prioritization frameworks for evaluating experiments. Almost none of these have direct ties to end user facing features so they can be hard to get a team interested in prioritizing them.

Typically this skill set only gets developed and internalized by a PM who is member of either a user acquisition or performance marketing team. Especially if they have been a PM for a dating, gaming, affiliate marketing, ecommerce or SAAS company. But once a PM has developed skill 2 they can join any organization and almost immediately increase the rate of experimentation.

libtechSkill Three: Rate of Idea Validation

Every company needs multiple PMs that have both skills above. In fact large economic value can be created by only having PMs with above levels of skills. But iconic product companies need to have at least one PM who can drive the rate of idea validation faster over time. The rate of idea validation is a measure how quickly new ideas are evaluated to be killed or further resourced. Some companies call this innovation while others call it R&D. Regardless of what they call it, all companies are a function of how well they allocate their resources (I know, so very clinical) as well as how big the market is for their products. Thus finding large new market opportunities as efficiently as possible is straightest path to create outsized value.

Great PMs consistently find great new markets to build new products for as well as totally rethink existing markets to challenge status quo. It is almost magical to see this process in action because it turns out that there is no conscious way to learn how to do it. Great PMs are great because they have “taste” in picking problems & coming up with solutions. Really fucking good taste. This almost intuitive sense of taste enables them to allocate resources more efficiently than everyone else. They make great decisions with partial information while under great pressure while avoiding dead ends.

I believe great taste can be developed but not in a linear manner that is predictable or time bound. The best, and perhaps only way, to develop great taste is to be interdisciplinary and to gather a large variety of life experiences to draw upon. This is why Steve Jobs’ focus on the intersection of technology and liberal arts has always made a lot of sense to me.

So, in a nutshell, Great PMs improve the following three things:

  1. Rate of execution
  2. Rate of experimentation
  3. Rate of idea validation

When you find a great PM that has developed all of the skills above give them whatever resources they need to launch a product. It will likely succeed beyond anyone’s wildest expectations.

Thanks to Andrew Chen and Semil Shah for reviewing an early version of this.

Update: Cleaned up some formating and improved the TL;DR based on feedback.

3 Types of Notifications

Photo credit: Post Memes via Flickr

Talking with founders is one of my favorite parts of being an early stage investor. Recently a founder casually mentioned there were three types of push notifications. I pounced on the comment as I had thought the same thing independently and was I excited to hear someone else’s definitions. Turns out we were in completely different places: the founder was thinking about notification technology while I was thinking notification use cases.

The three technical types she outlined were push, local, and in-app. In reviewing the iOS dev docs, one could also frame three as badge, message, and sound. The reality is that on either dominant mobile platform, notifications have a lot of unexplored potential in terms of tech and UX.  And as the sophistication of notifications design grows in support of user engagement (not new user growth as is commonly discussed) the need for specific language to analyze the different types of notifications grows since not all of them are made equal!

With that in mind I’m going to enumerate the key notification use cases given their importance to mobile application design, engagement, and growth. Personally, I’ve come to believe there are three common types of use cases for notifications*: (1) user-generated, (2) context generated, and  (3) system generated. Diving into each a bit more:

  1. User-generated notifications: These notifications contain content created by a human using the app to other humans. Generally, these are the most engaging but especially so when the content they contain is private and directed to specific people. Mobile messaging is the highest volume example of this type of notification but other examples include comments / likes / favorites on posted content or @ mentions. My current favorite example of this notification is getting a new photo of my son from my wife.
  2. Context-generated notifications. These notifications are generated by an application based on the permission of its users. This is the fast growing category of notifications because the amount of machine readable data mobile devices create: location, contacts, calendars, and much much more. The norms around context-generated notifications are still be worked out between developers and users. Location-based notifications currently dominate this category but other examples include information about your next meeting (time relevance) or updates about your favorite sports teams (interest relevance). My current favorite example of this notification is when I get notified there is a designer nearby via Highlight (disclosure: I’m an investor there).
  3. System-generated notifications. These notifications are generated by an app based on the needs of the app. This type of notification can usually be called re-engagement at best or spam at worst.  Sometimes these can create value for the end user like letting you know a friend has started using the app or that there is a sale on in app purchases.  Actually I can’t say that with a straight face as I don’t think they’ve ever created any value for me. Wait, I got one: security alert that someone has requested to reset my password. Well, until those started coming every few days anyhow. 😦

I’d like to pause here to solicit feedback on this classification and specifically to find examples that break it. After digesting feedback I’ll follow up with some more ideas on notification design.

*While I’m framing this in the context of mobile push notifications I think these categories work fine to classify notifications delivered via other channels including SMS, email and soon to arrive widely web browsers notifications.


Getting to use Facebook for work, while I was part of the company for five years, was hard to top but I found a way – playing mobile games is now part of my DFJ job description.

I’m thrilled to share my third investment for DFJ and my first out of our eleventh fund, into Kiwi’s Series B led by Northgate Capital.  Kiwi is an AAA mobile gaming company co-founded by Omar Siddiqui and Shvetank Jain and initially funded by Omar Hamoui and Sequoia Capital. This team is one of the rare Android-first developers that have gotten five of their titles into the top twenty-five grossing of the Google Play Store. You can read more about Kiwi’s Series B funding here, here and here.

Now, as has become a bit of a personal tradition, I’d like to share the story behind my investment into Kiwi.

Since joining DFJ just over a year ago I’ve been talking about the large opportunities available on Android.  As a result I’ve been lucky enough to see some of the most exciting Android-first focused startups that are pushing the mobile experience further than most folks in the Silicon Valley think is possible. Nevertheless, I had trouble finding right team, concept and investment opportunity to get behind and wasn’t sure if I’d find the right Android-first company.

It turns out, sometimes as an investor, you have to be found instead of doing the finding.  Lucky for me Omar Siddiqui found some of my blog posts about Android so he asked for an introduction via our mutual friend Matt Wyndowe.  Even though Matt spells his last name so wildly, he is anything but wild when it comes to making introductions. He is one of the most thoughtful super nodes I know!

In our first conversation, Omar captured my imagination with his vision for building an amazing Android-first company focused on AAA games and distribution. As I spent more time with him and the team I started to learn new things about mobile gaming and Android that floored me.  Especially around how Android distribution and monetization were working in the wild.  The scale of Kiwi along with the number of Android product experiments they’d made was astounding.  Just as importantly, Omar and I had a very natural communication rhythm that led to several deeply engaging conversations around the future of Android and mobile in general that profoundly affected my thinking. Needless to say, I quickly knew I wanted to invest.

The only rub with investing was that Kiwi already had an incredibly strong investment syndicate with Northgate committed to leading the Series B with participation of Sequoia and Omar Hamoui (the founder of Admob) who led the Series A and Seed rounds, respectively.  Not such a little rub, in truth. Luckily, Omar and his existing investors were equally excited about having DFJ participate so we came up with a way to make it work for all parties. This was also a particularly good learning experience for me since Kiwi was my first series B investment in addition to being a multiparty negotiation.

Right now, the Kiwi team is hard at work on its next generation of AAA games. And in seeing the internal demos I am confident that mobile gamers are going to be floored with what they deliver.  If you love producing world class AAA games please reach out to me as the Kiwi team is planning to hire another executive producer now that the series B money is “in the bank”.