Warren Buffett is known to have said “In business, I look for economic castles protected by unbreachable ‘moats.’”
Once one company finds product/market fit, competition becomes inevitable. Prices drop, features standardize, and it becomes harder and harder for the original company to differentiate. This is mostly theory of course, but many industries and markets exhibit this behavior over time. This is generally great for the consumer, but can be very hard on companies.
So who are the companies that succeed in the hyper-competitive free market economy? It seems to me they either:
1) Are exceptionally innovative (moving into new markets with ease)
2) Have found economic moats to keep the competition at bay (their products and markets intrinsically keep competition at bay)
(The third option, what the desperate companies do, is generally either unsavory activity like excessive litigation or illegal stuff like price-fixing cartels)
The internet marketplace carries with it a double-edged sword for entrepreneurs. On one side is access to a market of billions of people instantly, and on the other side is competition with hundreds if not thousands of competing services.
The best entrepreneurs on the web find a way to harness the former while managing the latter.
How do they do this? Can we create a framework for understanding which businesses can succeed in the hyper-competitive marketplace of internet business?
Here is my crack at the dynamics that make for these long-term, sustainable, successful businesses:
1) With valuable historical customer data
2) Access to the long tail
3) A propriety network with strong bonds between members
4) An unusually strong brand
I’ll explain each briefly and try to look at some of the successes and non-successes of some of the big companies within this framework.
1) With valuable historical customer data
Usage of a particular product or service almost always throws off data. The products or services that improve substantially with this data over time can create enormous difficulty for competing services. Competitors cannot simply compete on features or price, because the incumbent’s value to the consumer has had months or years of usage that improves that particular product. Ebay is a good example, as is Airbnb or my previous company Roomorama. If you have 100 positive transactions on Ebay with a 100% satisfaction rating, or hosted 30 stays with good reviews, that platform is immensely more valuable to you than a new competitor. And this value has built up over years in many cases, something that can be very hard for a competitor to replicate.
This does not only true of marketplaces, though that dynamic is frequently there for marketplaces. Companies like Mint.com or Netflix have the same effect. I have 5 years of personal finance data in Mint. Being able to query that data on my personal finances back 5 years is very valuable. Resetting my data with a new service probably won’t happen anytime soon. New services won’t have access to old bank/investment accounts, my categories and notes, etc. Companies with strong recommendation engines (in theory :) get better with usage as well, such as Netflix. A history of activity makes Netflix increasingly valuable to me over time.
2) Access to the long tail
Access to the long tail or marketplace liquidity is also a tremendously difficult competitive barrier to overcome. This is for two reasons: 1) its hard to aggregate the long tail and 2) the long tail grows on the end of the long tail, creating a snowball effect. This is true of most of the big marketplaces out there today. As consumers, we come back to marketplaces that help us find what we need, even if it is very unique to our needs. Then, a portion of us demand-side participants then become supply-side participants (typically extending the long tail of the marketplace in a virtuous cycle that makes the marketplace even harder to displace). Etsy is a good example of this, as are most of the big marketplaces out there, horizontal or vertical.
3) A proprietary network
Third is access to a proprietary network. Interpersonal networks, whether they are hobby related or professional related, are incredibly important to our lives, personally and professionally. These are hard to build, as they required multiple mutually beneficial interactions and time. But once they’re built, they’re very hard to replicate. This is true of Facebook, LinkedIn and Twitter, but also of small forum and community sites with active members (PocketFives for poker as an example relevant to me a few years ago). I believe this is true of Quibb. Yes, there are many places I can post what I am reading, but very few places do I have a history of interaction that I can find here.
4) A very strong brand
Good brands are an incredibly strong asset. It is why Coke has been a solid company for a long time. I’ve heard it said that the company actually outsources almost everything at this point, and is essentially a brand holding company at this point. A good brand is very hard to replicate. An example of a company that may not have any of the other moat dynamics but has this one is 37signals. Their software is very good, but not impossible to replicate. What is hard to replicate is the 37signals brand. Tom Tunguz wrote a post about this being something web companies will need to focus on going forward, and I could not agree more.
So if my theory is that these are the dynamics at work for successful, long-term businesses, what are some of the features of companies that may struggle with competitive pressure?
1) Arbitrage based businesses
Companies that rely on a short term CAC < LTV equation, but nothing else will struggle over time. These companies may look attractive in the short term, but CAC almost always goes up, and LTV almost always goes down. Groupon would be a good example of this. In the short term, they found a replicable business model that acquired customers for less than they were worth in revenue. But in the absence of the dynamics above, their CAC went up, their LTV went down (for reasons including but not limited to competitive pressure), and profits were squeezed immensely.
2) Businesses that do not increase in value with customer usage
Many pure SaaS businesses would fall into this category. There are many small startups that I would say fit into this category, and I don’t feel the need to call them out. But if you’re simply creating software that can be replicated with a handful of solid engineers, it will be a difficult road ahead. Especially if customers will pay a good price for your software, you will find new competitors every day and will likely end up in an unpleasant pricing/feature war.
3) Purely content based businesses
There are solid content brands out there, so there can certainly be exceptions. But the explosion in user-generated content and internet users with short attention spans, coupled with difficult monetization headwinds (CPMs still going down), can make that a tough business. Certainly not impossible, but just tough. Newspapers are probably the most notable example here.
So theres my personal framework for what can create long-term, large, sustainable web based businesses with strong economic moats.
There are exceptions, and by no means do I think this is definitive. But its a framework I might use if looking at potential investments. And after all, as entrepreneurs, we’re basically investors ourselves :)
I’m curious to hear what others think.
“I don’t think it’s coincidence that Cap and Daniel and Doug work in design, a discipline that at its best is grounded in empathy, but regardless of which field they happened to work in, they offer examples of exactly the serendipity and opportunity that can arise when we hold the door open, just a little bit, for that person entering behind us. All we have to do is not be afraid of who we’re letting in.”
The difficulty seems to be, not so much that we publish unduly in view of the extent and variety of present day interests, but rather that publication has been extended far beyond our present ability to make real use of the record. The summation of human experience is being expanded at a prodigious rate, and the means we use for threading through the consequent maze to the momentarily important item is the same as was used in the days of square-rigged ships.
“In many ways, the work of a critic is easy. We risk very little yet enjoy a position over those who offer up their work and their selves to our judgment. We thrive on negative criticism, which is fun to write and to read. But the bitter truth we critics must face, is that in the grand scheme of things, the average piece of junk is probably more meaningful than our criticism designating it so. But there are times when a critic truly risks something, and that is in the discovery and defense of the new. The world is often unkind to new talent, new creations, the new needs friends.”
– Anton Ego, Ratatouille
Just wanted to post a quick thought on two-sided markets. The above image demonstrates graphically what I mean.
In two-sided markets, there is generally an ideal ratio of buyers and sellers, and its important to know what that ratio is. The supply side of many of these markets (Ebay, Roomorama/Airbnb, Elance) is generally easier to acquire, because the cost to list is relatively low (list a t-shirt, a property or your consulting skills). Generally the cost to acquire those customers is linear (ie it costs $20 to acquire a new supplier).
From the demand side, the value of the marketplace is determined by the quality and quantity of listings. If there is only one property in Boston, those coming to look for properties probably wont find what they need. If there are 300, theres a very good chance I’ll find what I want. But if there are 200, its probably not all that less likely I’ll find what I want. So acquiring those extra 100 properties is a waste of time and money.
The value to the demand side follows an S-curve. Value increases with a higher supply:demand ratio, and then it reaches some peak value at which it levels off with an increase in that ratio.
It is important for those running two sided marketplaces to figure out that ideal ratio (the shaded part of the graph). Is it 7:1? 30:1? Underserving or overserving the demand side can be very costly, and making sure you’re in that sweet spot is crucial to making your marketplace work.
I think its a great time to buy Google stock, and heres why:
1) Incredible Business Model
Search advertising is still the most effective advertising medium for companies of almost any size. Purchase intent is the holy grail for advertisers, and they have this nailed. And if you think display is the future, well they have a ton of display inventory too.
2) Great Founder Leader
Its typically good to have someone running the show that has been a part of growing the culture for a long time. And by all accounts, Larry Page is a very capable CEO.
3) Domination of a Huge Market (>80% global market share)
Despite many trying to unseat Google due to the attractiveness of its business, few are very good at it. They’ve maintained search marketshare in the face of a handful of new competitors in the last few years.
4) Leaders in So Many Fundamental Web Products
Chrome, Gmail, Youtube, and many others. They are the default place for so many fundamental web activities.
5) Solid Mobile Foundation
While no iOS (imo :), Android is solid. And they clearly see mobile advertising as an important ad product as well, and are pushing that product pretty successfully as well.
6) Infancy of New Products (G+, Google Drive, Google Docs, Google Fiber?)
They’re constantly developing cool new things. Say what you will about G+’s usership, but its a pretty cool product. It may take time, but I think G+ can only go up from here.
7) Renewed Dedication to Making Money
They’re starting to charge more for their business apps, and Larry Page is very focused on not just creating great products, but continuously improving margins and financial results.
Just wanted to write a short thing about the recent tragedy in Newton, CT.
I don’t feel the need to go into gun control or mental health. Both of these are not debatable. We need action on both, and to me, they’re so self-obvious that even debating them discredits that fact. (This doesn’t mean I don’t think they should be debated, just that I’m done doing it, because I have found very few of those conversations reasonable).
Anyways, it may seem broad, obvious, and overly simplistic:
But we need to start being better to people.
People are “different”, and “weird” and socially distant and awkward. And it seems to me that those who have been applied those stigmas are increasingly isolated. It probably has something to do with social media.
Like all of us, these kids are looking for acceptance, love and a sense of community belonging.
We must seek to provide that, especially when we see it lacking. It starts with a smile, saying hello, and asking how their day is going. These little things go a long way.
This kid was evil and what he did was evil. And I’m not saying what he did could have been necessarily prevented. Just that I think we need to remind ourselves to be good to people, no matter who they are, because it requires so little of us to do so.
Knowledge sharing on the internet can do better. Most learning online has come in two forms, written/static content and more recently, video. Both miss the mark on a key aspect to efficient knowledge transfer: conversation. Learning is mostly an iterative process: what do I know?, what do you know?, what is the context of what is being learned?, how does the information solve my problem?, what is my problem? A back and forth around these questions bring about the most efficient transfer of knowledge. A recent article in Techcrunch is relevant here: http://techcrunch.com/2012/12/08/the-underachiever/. The writer presents the dichotomy as “didactic” vs. “dialetic”. People prefer conversation to preaching.
And as the internet moves from static to interactive (websites vs. apps broadly), its as good a time as ever to move towards a more interactive learning environment, one that promotes conversation over one-way lecturing, allowing for a more effective spread of ideas.
The 21st century worker must have a depth AND breadth of knowledge. Where traditional education and industry experience can help in former, I believe a new system can help in the latter. The most valuable people in the 21st century economy will be the ones that can be valuable in a variety of ways, able to access knowledge easily and apply it to their specific context.
Bringing together two long-time central functions of the internet, the building of highly liquid marketplaces and the easy dissemination of ideas, I believe it is the right time a better tool to bridge knowledge gaps in a variety of contexts.
Much has been written recently about the Series A crunch. Seed funded startups are finding it difficult to raise a Series A and will have to either find an acquihiror or simply shut down.
This is neither “good” nor “bad” in any broad sense, but just a reality of capitalism.
But one thing that I think has been missing from the conversation is how good the explosion of startups in the last few years (which has precipitated this Series A crunch) has been for the average consumer of tech and internet based products.
Part of the reason that many of these companies will have to fail is that they cannot weather the difficult fund-raising environment with revenue. Many have gone the route of scale before revenue, and have just not been able to either sufficiently scale or sufficiently monetize.
These startups are creating tremendous value, and capturing very little. And for that reason, they are not sustainable businesses, and is part of the reason for pretty subpar VC returns in the last 10 years.
But lets think about what that means for the person consuming these services. Every dollar of value not captured by the business is a dollar of value “captured” by the consumer.
Its an amazing time to be a consumer of web services and tech products. Much of the internet is free or really cheap. Even paid services like Netflix and Spotify are incredibly cheap when you think about the value to a consumer. These companies may struggle financially (along with all of the other smaller startups that are now failing facing the Series A crunch), but they are tremendously value creating for those who consume them.
In the spectrum of value creators vs. value capturers (see Chris Dixons post on Builders vs. Extractors here: http://cdixon.org/2010/06/19/builders-and-extractors/), entrepreneurs are way on the side of creating more value than they capture (Wall St. being substantially more on the other side).
So if you have a friend at a startup thats not going to make it, at least thank them for putting themselves out there for creating something that has, even in a small way, added a whole lot of value to humanity, and required very little back in return.
The digitally connected world has allowed for the collection of an incredible about of data.
What’s cool about this from an Analytics and BI perspective is that we now have datasets that can lead to statistically significant findings.
The “long tail” of data (e.g. a very granular customer segment) is now open to insight, as we now have enough data to draw statistically significant conclusions.
Right now, the majority of the work in “Big Data” is being done on the infrastructure level. How do we collect this data in an efficient and scalable manner? How do we build an system for analysis that is fast and accurate? These are questions that companies like Cloudera (a company commercializing the open source Apache Hadoop project) are working on, and getting a lot of money from VCs to do it.
If the hard part at first is collecting, managing and cleaning the data, the exciting part is the analysis layer. Without value creation from data, its still just mostly an academic exercise.
As I see it, there are a few key analytical processes we undertake with data:
1) We analyze against our success metrics (basic statistical analysis to determine performance)
2) We run controlled experiments (causations, AB, traditional hypothesis testing)
3) We look for relationships (correlations, regressions)
4) We try to make predictions about unknowns (algorithms, machine learning)
5) We look for anomalies (outliers)
I still think we’re in the very early innings of this analytical layer. Traditional BI tools help us with #1, to a certain extent. A/B testing tools help us with #2. #3 still requires a bit of work and mathematical knowhow. #4 mostly requires some relatively heavy math/programming skills (however, there is a really awesome site called bigml.com that can create pretty robust predictive models, but that is also in the early stages).
I’m personally excited about #5 as I haven’t seen companies focus on this yet, and I think it can bring tremendous value. Anomaly detection can do two things. First, it can help troubleshoot data collection problems (not that exciting). But second, it can help us find true insights by focusing our attention. If we see most paid traffic sources generating a 5% conversion rate and 8 pages per visit, but a new one that generates a 20% conversion and 18 pages per visit, this is an outlier. Its a potential insight, and could theoretically bring substantial value to the organization (i.e. why is this channel bringing more successful traffic, and what other channels are similar).
Detecting anomalies is much more difficult that this simple example would suggest. Knowing expectations or classifications is a difficult step one before finding what doesn’t fit into those expectations. So I think we’re still some time off, but I think this could be a huge place for value creation going forward.
In any case, it will be interesting to see the “big data” world start to move down the value chain and to finally realize its fundamental promise, to bring to light otherwise hidden insights.
I recently just started using Instagram, and it had me thinking about why the acquisition occurred and was so substantial, what works so well about the product/network, and what it means for the rest of the social web.
It occurred to me that because Instagram’s network centers around a very specific behavior/media type (photos), the implicit relationship between me and my followers is fundamentally different than that on Facebook. There is an understanding that I will be sharing photos (potentially many of them), good or bad, and my followers have chosen to have them in their feed. If my photos start to suck, they can unfollow me, and while it may be a slight hit to my pride :), its not a big deal. They may still follow me on Quora for Q&A, or on Quibb for professional links, or on Foursquare for location updates.
I’d say that the proper analogy is the way in which Craigslist is being disrupted by startups attacking specific verticals, and focusing their own product on the specifics of that vertical, ultimately creating a better experience than the catch-all network itself can provide. And as the volume of web activity grows (mobile being a huge driver), I believe networks that are catered to unique behaviors will offer a much better experience for both producers and consumers of that type of content.
One can look at a variety of online behaviors that are actually better suited to unique networks:
Product sharing: Pinterest
Music: Soundcloud? Spotify
Status Updates: Twitter (sort of)
Short Blog Posts: Tumblr (sort of)
Links: Quibb? :)
To summarize, I believe these types of networks have a few advantages:
- They can focus the product on the features that are unique to that specific behavior
- There is an implicit understanding between connections about the type of content being shared (reducing reluctance to overshare)
- They do not force a reciprocal relationship (I can follow you, but you don’t have to follow me). This is a profound difference in how the network is constructed
These elements provide a substantial opportunity in the social/consumer web. Some are somewhat solved already, some are still TBD. It remains to be seen which behaviors can viably support a unique network, and if so, who will win. Competition is heating up in some of them (links, conversations), and some are yet to be seriously looked into (debates?, polling?) We are sure to see a mix of M&A and internal product development from Facebook to counter these threats.
In any case, competition in the social web is certain to heat up, and it will certainly be very interesting to see how it shakes out.
In light of a lot of discussion in the past couple years about learning to code in the tech/startup world, I wanted to write a brief post about the things that go into learning any new skill-set.
Fundamentally, once you have these few things down, I think you’re pretty far up the curve, and it then takes experience to get really good.
For many of us, getting this type of experience is probably not feasible (10,000 hours according to Malcolm Gladwell), and may actually be a poor use of your time and energy, assuming you have strong experience in other areas that provide tangible value to a team.
So here are the four things I think we CAN all learn, in any new field:
1) The scope
2) The theory
3) The syntax
3) The tools
What I mean by these four is the following:
1) The scope
The first difficulty in understanding a skill-set is to know what the total universe of possible problems and solutions someone with that skill-set might encounter. What are the different tasks that an engineer, or a designer, or a marketer might undertake on a daily basis? How do they approach those problems? How long does it take to solve them?
2) The theory
What are the basic underlying principles that that skill-set depends upon? In my limited experience with coding, there are a few things that I see are crucially important from a theoretical standpoint; proper organization and clear construction of variables and functions, an understanding of notions around recursion and iterations, and how to leverage conditionals. That isn’t all, and I’m going to continue trying to understand all of the basic theories around developing software until I understand them from a high level.
When I decided to take a month to learn about search engine marketing, first it started with a theoretical understanding of how search engines work. Its really important because then you can use that foundation to attack any new discrete problem or question.
3) The syntax
All skills generally have a language that just takes time to learn. If you don’t know the words, terms and acronyms that are used, you’ll have a hard time understanding what’s going on.
4) The tools
Tools are a practitioners best friend. They simplify complex tasks and help a person leverage other people’s great work in the field. Without them, you will be spending a lot of time doing work that others have already done. Learning math? You should know the basic formulas. These are the tools that can help you solve more complicated problems later.
I think its really important, especially for generalists, managers or C-levels to really understand these four things in all major competencies in which your company operates.
For many of us in startups that means engineering, design, marketing, finance, and then the particulars of your specific vertical.
I’m a long way from knowing these 4 things for each, but find that its all within reach, all of the help I need is out there in this wonderful thing we call the internet, and fundamentally, while experience is required for true skill, its not necessary for getting up the curve very quickly.