The Blog Of Matt Brezina Header Image

Secrets of Rapid App Development

I’ve given the talk embedded below at the First Round Capital CEO Summit as well as at the Lean Startup Conference.

The transition from the web to mobile has been dramatic for product development teams and the entrepreneurs that run them. We got comfortable in a web world where we could ship new code 24 hours a day, where analytics was made easy with a few lines of code, and where paid user acquisition channels were well understood. Mobile changed everything.

This presentation contains 10 tricks for how we’ve addressed the first of these problems – rapid development and testing of mobile software. Enjoy!

Hiring via API

It is easy for a team of developers to start a startup right now. Indeed, many are. There are now over a hundred developers in each new Y Combinator class! These are people that could otherwise be working at your company.

Why would someone start their own company rather than joining yours?  It isn’t because they are likely to make more money.   A developer has a much higher expected return going to work for Google, Dropbox or Square than doing their own thing. I believe it is culture, not money that is driving their behavior.

“Do my own thing.”  
I speak with lots of developers every week. If they are good I try to hire them. If I fail it is rarely because they choose another startup. Usually they say “I want to do my own thing.”  This translates to “I want to decide with who, how much, and on what I work.” This is a culture decision. This may be working only 20 hours a week. This may be working from 8pm-8am wearing PJs.  This may be working while traveling in Asia.

If there is all this talent out there “doing their own thing,” how does a company capture their value?
The simple answer is to become their customer. And no, i’m not talking about contract work.

I’ve never met a talented person who said “I’d rather do contract work.”  Developers don’t like equating their time to dollars.  They don’t like working on someone else’s stuff.  Good developers want to be as close to their customer as possible.

Partnerships written in code
Luckily APIs are an innovation that allow developers to create meterable value for a company without being employed by said company.  A developer gets paid for the value they create not the hours they put in.

Instead of building a notification system for your iPhone app (requiring an additional engineer and $150k in total costs), it is easier to pay Urban Airship a fraction of a penny every time you send a notification.

I encourage developer teams to make a meterable service that many companies need but don’t want to build themselves.

As a company scales they may end these API partnerships and vertically integrate – building every piece of their service and value chain. At 2M notifications per day, it might make sense to build your own Urban Airship. But maybe you don’t ever want to get into the business of sending notifications, and you don’t have to.

It is often said the best path to success is focusing on what you are good at. Everything else you should outsource to a partner.

APIs are partnerships written in code

Case Study: Sincerely Inc.
At Sincerely, many aspects of our service and business have been written by people who don’t work for us:
Postmark – instead of running our own SMTP server.
Appirater – instead of writing our own app review notification system
Hoptoad – instead of building our own crash reporting system
Urban Airship – instead of buildling our own notification system
Geckoboard – instead of making our own dashboard
Linode – instead of owning and running our own servers
Flurry – instead of writing our own analytics
Instagram – instead of creating our own community of photo content
Github – instead of hosting our own repositories

How you can hire us
Sincerely’s focus is making it easy to send real photos in the mail from a mobile phone.  We’ve gotten good at collecting physical addresses from users. We’ve learned how to bill users for small transactions.  We’ve set up relationships with print centers around the world.  We’ve built powerful customer support systems that address the challenges of sending real things in the real world.

There seems to be a new photo app launching every week. It would be a huge undertaking for each of them to recreate what Sincerely has built.  Our Ship Library for iOS makes it easy for any app to add photo print & delivery functionality in less than 30 minutes.
 
I believe we’ll see a pattern of more companies launching APIs and launching them earlier in their life.  It is the first step of a biz dev strategy – and it is a step that developers are comfortable with.  We get to talk to each other with snippets of code.  We don’t have go through a big legal process to work together. We can even create value together without having money change hands.  Heck, you can scale to hundreds of partners without having a biz dev guy – sounds like something a lot of developers I know would really like to hear.

What services do you wish developers “doing their own thing” would build?

*There is a good discussion going on over at Hacker News – join in

YC: The new grad school

I’ve always been interested in starting companies. Ten years ago I was an undergrad at Penn State. I was told by professors, my parents & the press that the best way to start a company was to go to graduate school, become an expert in your field and start a company with a classmate.   Ten years ago grad school probably was the best place to meet potential co-founders. That’s what Larry and Sergey did.  That is how Filo and Jerry met.

So I went to grad school.

I did not love my time stuck in the dark corners of the control systems lab at the University of Maryland.  The opportunities to evaluate and work with potential co-founders were lacking. Note* I realize this concentration would have been higher at a place like Stanford or MIT. Luckily I met Adam Smith on Craigslist during a summer internship, we became roommates, did a lot of brainstorming over hot pocket dinners and eventually started Xobni together.

From the outside Y Combinator may simply look like a new disruptive venture fund. And it is. But the secondary effects may be even larger.  

I often say Paul Graham doesn’t do YC for the money, he does it to teach.  He is a professor in a new type of university. And as an entrepreneur himself, of course it is a university he founded.

The practicum is launching your first product.
The Tuesday dinners are the weekly lectures.
The coursework is Hacker News.
Your fraternity brothers are the YC alums
And unlike contributions to your university’s general fund – when you invest in Y Combinator you get a return.

As a member of the 3rd ever YC class (summer 2006) I had a sense that I was getting my chance to be an early alumni of a prestigious new community – not unlike the early graduates of Harvard.  As the years have passed, as Y Combinator has grown, this has only become more true.

Founding Sincerely 

I co-founded Sincerely with a former YC classmate, former Xobni employee, and friend, Bryan Kennedy.  Bryan’s YC-backed company Pairwise was an early hit, but never turned into the business he wanted it to be.  Bryan is the quintessential internet hacker; he has been building websites and web communities since high school. He is part developer, part designer.  I recognized his talent early during YC and when Bryan shut down Pairwise I grabbed him to run Xobni’s web team as employee #4.  Later Bryan and I worked side by side running a very analytical growth team at Xobni.  After several great years with Xobni Bryan got excited about the iPhone, left Xobni, started Swipe (a credit card processing app), and sold it within 8 months on the heels of Square’s launch. Now Bryan and I have co-founded Sincerely together – we’re making it easy to send real printed photos from your mobile phone. We just launched our first product Postagram earlier this week

Lesson learned  

YC founders, look around you. This is the greatest concentration of company & product building talent you are likely to experience in your life.  There are potential employees and co-founders sitting around you at each dinner.  Help each other succeed. And who knows who you’ll end up starting your next company with.  

Welcome to the new grad school.  

Sincerely Inc. My new startup!

Sincerely_Logo

I’m extremely excited to announce my new company Sincerely today. We are focused on making it easy to send real physical photos from your mobile phone. Simple vision. Huge market.

I’m very excited to be working on this company with former YC classmate and Xobni teammate Bryan Kennedy. I’m also blessed to be working with some other really, smart guys.

We believe that a printed photo is the most ubiquitously appreciated gift in the world.

We believe that in an increasingly digital world, physical matters more than ever.

We believe the iPhone4 changed everything. It is the new and future of point and shoot cameras.

We believe it should be as easy to share a physical photo as it is to post a photo to Facebook.

We believe that as we build Sincerely Inc. we will make it easier for more people to be thoughtful.

I believe this is going to be a lot of fun

We launched our first product today. Postagram: easily send an Instagram print and a short note in the mail for 99 cents. Check it out here: http://www.postagramapp.com

To learn about future Sincerely products, sign-up at http://www.sincerely.com

Startup Success Demands Close Collaboration – Revisited 3.5 years later

I originally published the following blog post on the Xobni company blog on September 18, 2006.

That was over 3 and half years ago. Xobni was still based in Cambridge, MA, my cofounder and I were still living on hot pockets and chips ahoy cookies, and we were sharing a bedroom in a friend’s apartment. It was during this time that I crystallized what is in my mind one of the most important truisms of founding a company; starting a company is like getting married and having a kid. Your cofounder becomes your spouse and your startup becomes your baby. You and your cofounder will learn to know each other just like an married couple. You’ll fight like a married couple. You’ll share successes like a married couple. And you’ll both care deeply for your startup, even though you may differ on how to help it succeed.

And now, that original blog post:

Startup Success Demands Close Collaboration (Sept 18, 2006)

People say that the most important decision you will make when starting a company is choosing a cofounder. It bears repeating. The commitment you should have to your cofounder is not unlike the commitment you make when you get married.

Today, while we were sitting at Arlington Diner eating our 3pm breakfast, we came up with five questions you should ask yourself before getting married. These questions are also relevant when starting a company with someone.

1. Are you great at things your cofounder isn’t?
It is important that cofounders exhibit complementary skills and common objectives. Identify your weaknesses and make sure those are your cofounder’s strengths. I want to spend most of our effort on perfecting the presentation layer of our first product, but Adam makes sure we invest in our backend code so it is easily expandable to future product lines. Adam is good at Adam stuff; I’m good at Matt stuff; together we are good at Xobni stuff.

2. Can you commit to a 2 year cell phone contract with a “family plan?”
Sorry Mom, but Adam and I just signed up for a family plan. There is something very permanent about signing a 2 year contract with someone you are not related to. The funny thing is Adam and I didn’t even hesitate because Xobni has many years ahead of it.

Side Note: We signed up for Sprint with the EVDO data plan. It is awesome. We each got a Treo 700wx which runs Windows Mobile. There were two motivations behind getting these smart phones. First, it is much more relaxing being away from your desk for an hour when you know you have access to all of your email anywhere you go. Second, many of our users are adopting these devices; it is important for us to understand this trend.

3. Can you share a bedroom?
If you can’t share a bedroom with your cofounder for a month, you probably aren’t right for each other.
Adam and I are currently sharing a room. Adam remarked the other day that it feels like summer camp. It reminds me of our first week back in the MIT dorms. This is not our permanent living situation and we appreciate that a degree of separation is healthy. However, we are temporarily living with our friends until we decide on a permanent location for our growing company. Sharing a bedroom is a temporary sacrifice for ultimately making the best decision for the company.

4. Do you trust them with a joint bank account?
Adam and I aren’t taking individual salaries. We put enough money into a joint personal account to pay for things like rent, food, and Adam’s voracious appetite for Edy’s ice cream and Chips Ahoy Chewy Chocolate Chip Cookies. My personal indulgence is the recently discovered bliss of Bustello coffee. If you can’t trust your cofounder with your opulent salary of $1,200 per month, you are going to have trouble when the company bank account has more than three zeros.

5. Will you make dinner AND clean up afterwards if your cofounder is drowning in serialization code?
I don’t plan on marrying someone that even knows what serialization code is, but the analogy is valid. You are both going to encounter difficult challenges and it helps to have someone to lean on when stressful times arise.

I can’t think of any point in my life that will be more conducive to this type of commitment than right now. Remember, you are not just getting married, you are raising a child. We called our baby Xobni. Next installment: Xobni learns to walk.

Xobni’s 5 stages of growth & 5 pivots (preso from web 2.0 expo)

I recently had the pleasure of presenting twice with the leaders of the lean startup movement Steve Blank and Eric Ries. They’ve started a startup movement I’m a very big fan of. If you aren’t familiar with their work, I’d recommend reading their blogs. To concisely describe this movement we’re calling “lean startup” I’d liken it to a method of company building we’ve subscribed to at Xobni for a long time and that has only recently had a name. A Lean Startup is a startup with quickly iterating cycles of customer development and product development. Lean Startups stress product market fit before scaling and are characterized by a metrics driven culture and low cost experiments. I love this stuff. I can’t imagine building a company another way.

Today I spoke about Xobni’s lean startup tactics as part of the O’Reilly’s web 2.0 expo. The slides are below.

Freemium ain’t new. It is just a new word.

I recently spoke at the Freemium Summit in San Francisco.  The speaker line-up was full of great companies executing on freemium models in the consumer and enterprise space:  companies like Dropbox, Automattic,  and MailChimp.  I was asked to speak as part of the Freemium for Enterprise portion of the program.

Below are my slides from the presentation.  I split the presentation into three sections.

  1. First, an exploration of the freemium business model currently being applied at 3 public companies: LogMeIn, eFax (j2 communications) and Skype.  In studying the public SEC documents for these companies we’ve found a consistant 10 to 1 ratio between free users and paid users. A goal we’re trying to achieve at Xobni.
  2. Second, I show six business/market conditions that can suggest implementing a freemium model.   Many of these conditions are derived from discussions I’ve had with Sean Ellis, who blogs here and has had great freemium experience working with companies like Logmein, Xobni and Dropbox.
  3. Finally, any time I give a talk at a conference I try to give concrete, actionable examples of lessons learned from my own experience in building Xobni’s business.  In this presentation I’ve dedicated the latter half of the content to specific successes we’ve had at Xobni. And this was all made under the warning derived from Paul Buchheit’s brilliant quote: advice=limited experience + overgeneralizations.  So be forewarned :)

And one clarification on a slide that will make no sense unless explained.

golden gate park freemium business

Slide 4  This is a shot of golden gate park

  • My point is that golden gate park is actually a 120 year old freemium model.  90% of visitors don’t pay anything to enjoy the park.  10% pay for the premium services like the tea garden, conservatory and golf course.
  • The foot traffic of free visitors leads to more sales and admissions for the premium services of the park.
  • I looked at golden gate’s financial documents it appears that they still aren’t profitable.  Premium services only provide 40% of the budget.  So, like most things in government, it isn’t profitable (or even break even), but it is at least an attempt at a freemium business.

So with that, here are the slides:

Social Networks, The Monkey Sphere, and Moore’s Law of Human Relationships

Just before Christmas I did an interview with the SF Chronicle for an article titled  Year in Review: Social networks come of age by Benny Evangalista.  Benny and I had a conversation about social networks and how far they have come. In the article he conjectures that social networks are here to stay, as is evident by their adoption inside big companies.  Our discussion revived a topic I’ve been wanting to write about after several years of working in this space; social networks represent the next step in expanding the capacity of human relationships.

Telephone, Email, Social Networks and the Monkey Sphere

The Monkey Sphere (also known as Dunbar’s number )  is a theory from evolutionary biology which was derived from the study of groups of monkeys (or more specifically non-human primates) in Africa.  Researchers studying monkey clans found that as a clan’s size grew towards a magical number of 150 monkeys, communication systems would break down, and the clan would break into two separate clans.  These clans would then each again grow to 150 members, split into two groups, rinse and repeat.  Researchers postulate that this barrier of 150 members could not be exceeded due to the communication strain on the clan as the number of nodes exceeded 150.  Dunbar and others postulate this same theory applies to human relationships.

cavemanFor most of human history our interactions were limited to a small social group, generally limited by physical geography.  Much like our primate relatives, this led our ancestors to only maintain relationships with those that they could talk to in person on a regular basis.

Enter the telephone

telephoneThe physical barrier was broken when the telephone emerged. We were no longer required physical proximity to another human to maintain a productive relationship (I’m ignoring physical letters because of their slow communication cycle time).  With the introduction of the telephone our monkey sphere increased monotonically.  But telephones had a limited effect because communication was limited to one on one discussions.

Email

emailNearly one century later the monkey sphere was stretched further by the internet, and specifically email.  Not only could communication happen without the need for physical proximity, but it had less marginal overhead (no small talk), and it was our first of the democratized one to many communication mediums (newspapers, books, and magazines only provided one to many communication capabilities to a limited subset of the population).  Oh, and to add to the argument, email is asynchronous – you don’t  need to participate in the conversation at the same time as the other members.  Our monkey sphere started expanding geometrically.

Social Networks

twitter-facebookBack to my discussion with Benny.  The topic at hand was social networks and how they represent the latest step in expanding our monkey sphere beyond natural levels.  Social networks have no need for physical proximity, they have less marginal overhead than email, they are asynchronous, they are one to many, and increasingly they are public, not even requiring that you know the people you are communicating with (ex: Twitter and the increasingly public news feed on Facebook).  Social networks have changed how, with who, and how often communication is created and consumed.

The telephone, email, and social networks all represent big steps in our communication evolution, and it has all happened in the last 100 years.  Interestingly, it also appears to be accelerating.  I believe that I currently have personal relationships (which I define as knowing their name, face, where they live, what they do, and having shared some experiences together) with nearly 1,000 people.  What next step in communication technology will take me to 10,000 personal relationships? Is it possible?  The communication of our location might give us the next big leap.

10,000 personal relationships might sound ridiculous. But so did a 2 GHz processor in 1988 when I was booting from a 5 1/4 inch floppy disk so I could play video games on my dad’s 10MHz Digital computer. Is it possible there is a Moore’s law for human relationships? I think it is more likely that a Moore’s law exists for human relationships than a Dunbar number.

Moore's Law

Moore's Law

Follow the Sales Guys

Want to build the next big personal productivity application, communication platform or X?  Follow those that have the most to gain from it.

With any new technology, the users on the forward edge of a trend are the people who stand to benefit most.

Follow the sales guys

Those that benefit most from improvements in personal productivity are sales people. Sales jobs have the clearest correlation between personal output and compensation among any role inside a company.  This direct correlation leads to millions of sales people across the web searching for any edge they can find to increase their sales (and therefore their compensation).  Make twice as many calls in one day – make twice as many sales.

Jawbone, Blackberry, Xobni, Virgin America’s WiFi enabled flights – all of these products helped sales people be more productive.

I like building stuff for sales people because of how clear their incentives are.  To understand a man, understand his incentives.

Follow the marketers

The people that benefit most from improvements in communication technology are marketers.  In no other role is success more closely tied to getting your message heard by more people.  Get your message heard by twice as many people – drive twice as many visitors to your product/website/store.

Twitter, email, Facebook – whether it be marketing a product or marketing themselves, the people that have the most to gain from new communication technologies are marketers.

Want to build a new startup in the personal productivity space? Listen to the sales guys.  Want to build the next big communication platform?  Get to know some marketers.

I guess this is all just a corollary of knowing your customer.  But I thought it was important to break it down just the same.

What do you want build?  Who are you following?

The 4 Metrics of User Acquisition and the Customer Bulls Eye

To the outsider it may appear that 100% of Xobni’s engineering effort is focused on the development of our Outlook plugin. Surprisingly, this is far from true.  Anybody who has operated a software company will tell you there is a lot of engineering effort required to turn a product into a business: billing infrastructure, bug reporting, user tracking, etc.  I suspect over 20% of our engineering effort is focused on business engineering.

Behind Xobni the product is Xobni Corp. the metrics driven business – rich in engineering complexity.   I personally find this hidden Xobni almost as interesting as the core Xobni product – especially at this stage of the company, the stage I’ve previously called “engineered growth.”

One result of two engineers founding a company is that the engineering culture will bleed into every nook and cranny of the company’s culture.  Recently I was talking to Tristan Harris, the founder of Apture , about company culture.  He said the competitive advantage of the world’s greatest companies is a specific and overriding culture that is pervasive in everything they do.  Tristan previously worked at Apple and said their culture was design.  We decided Google’s is speed and Zappos is customer service.  He asked me what I thought Xobni’s culture was, and I said “measurement.”  This culture exists in everything we do, including the business aspects of building xobni – specifically driving user acquisition and generating revenue.

I want to share some glimpses into our user acquisition and monetization machine.  This is the stuff you’ll never read on our company blog or see in the press coverage.

4 Metrics Of User Acquisition

For any user acquisition channel – PR,virality, advertising, search engine marketing – we measure success on four key metrics:  engagement, virality, profitability & scalability. I visualize each customer acquisition channel as a dart on a dartboard.   Our goal is to optimize those channels so they each hit the bull’s eye.  We are also always searching for more channels to throw at our dart board.

CustomerBullseye

We’ve made systems and dashboards to track and a/b test our user acquisition funnel.  We do this for each of our user acquisition channels.

For each channel we can A/B test over 50 landing pages.  Then for each channel and landing page combination we can see how many installs we got and what the conversion rate was to install (Scale), the percentage of users active in the last 7 and last 30 days (Engagement), how many new users these installs generated through our in-application invitation systems (Virality),  what percentage of these users converted to Xobni Plus and how much we paid for one of these users (Profitability), along with several other intermediary steps to each of these objectives.

We can take several groups of users (all from the same channel) and show them different messaging on a landing page, and see the resulting effects on scale, virality, engagement, and profitability.  Or, we can run one channel at a loss simply because those users are highly viral.  This reminds me of my optimal control research from grad school, except instead of optimizing on robot speed, accuracy and energy consumption, we are optimizing on scale, virality, engagement, and profitability.

1. Virality

We are adding more and more methods by which users are telling each other about Xobni through product interactions.  Right now we only track two.  The next generation of our tracking system will have over 10 viral inputs.

2. Engagement

We record other engagement metrics beyond the active user count – like clicks per user per day, searches per users per day, etc – however right now we review these metrics in one-off manner.  I’d love to add them to our dashboard.  It’d be fascinating (and useful) to know that a user acquired with a banner ad on Facebook make 5 times as many clicks/day on our premium upsells as a user acquired through twitter.  And, I’d love to know that I can increase the number of clicks to 7 if I send these users through a landing page that messages heavily on our premium product’s features.

3. Profitability

As previously discussed, Xobni Plus is just one of 5 revenue streams that will be driving our business by the end of 2010.  We’ve already announced Xobni with Salesforce and Xobni for Blackberry.  We’ll need to add these new revenue driving products to our dashboard and I’m excited to have the new ARPU (Average Revenue Per User) to play with.

4. Scale

This is where the creativity of an artist and the tenacity of a hunter intersect.  We are always searching for new darts to throw at our dartboard.

Sound Interesting?

If this work sounds interesting to you, you are in luck.  We are currently looking for a strong software engineer to own the fidelity, accuracy, display and availability of these dashboards and build the next generation of systems that will allow Xobni to make product/company strategy decisions based on data – our preferred method of making all decisions.  The ideal candidate should have a passion for customer development and the metrics of user acquisition and monetization.  The ideal candidate has probably read much of the writings of Eric Ries, Sean Ellis, Hiten Shah & others.  They should have a strong grasp of php, html & mysql – while these are the current technologies used in our internal dashboards we are always open to using the best technologies and platforms that will accelerate our steps toward making decisions faster.  If you are interested in this stuff and want to discuss opportunities for joining Xobni, check out this job description and email me at matt at xobni