It is amazing how few internet execs or entrepreneurs make systematic use of the data available to them. Sometimes one just wants to scream out: “You are on the internet, you have data, lots of data – use it.” Make use of your data and you will have a far more successful web site.
Here we will discuss two aspects of internet business model design closely linked to data: Analytics and experimentation. They actually go together; the ability to measure and the ability to test are a powerful combination. Obviously, even on the internet, there are some things you cannot try out without risking your current business. But much more experimentation is possible than people realize. Two companies that have engrained analytics and experimentation into their DNA are Zynga and Netflix.
Ken Rudin is General Manager of Analytics at the social media game company Zynga. His whole career has revolved around analytics – but this is his dream job. Imagine the possibilities to measure and test Zynga has with its 100m monthly players. Rudin said about Zynga: “I’ve never seen a company that is so analytically driven. Sometimes I think we are an analytics company masquerading as a gaming company. Everything is run by the numbers.”
In a presentation Rudin held before an association (TDWI, The Data Warehousing Institute) in late 2010, Rudin outlined three development stages of analytics: Reporting, Analysis and what Rudin calls Impact Game Design. All companies report, it is necessary but usually does not lead to increased value. Analysis will result in a little more insight, but in combination with experimentation it becomes very powerful. Impact Game Design actually is the combination of analysis and experimentation and for Rudin the pinnacle stage of development. Here, we will talk about Impact Game Design as well as another form of combined analysis and experimentation, A/B Testing.
Take, for example, advertising. For testing out different web page setups, for example a set of pages with advertising and one set without, A/B Testing is perfect. In the real world, it is difficult to offer advertising to one set of your customers and leave it out for the other – and then analyze the results after a month to see the impact. Even if you have the luxury of running your business in different locations and you can, say, run the advertising trial in Denver and not in Detroit, you will not come close to generating the amount of data that an internet trial would provide you with. The length of time people spent on your web site, what they clicked at, what they avoided, what they bought.
Zynga uses “Impact Game Design.” It is much more comprehensive than A/B Testing. As Rudin describes it, Zynga hurls many different games out at the community, and “most will fail.” The products are created with just the amount of effort required to create a working game, but not more. These “Minimum Viable Products” (MVPs) are buggy, but yield the required insight.
Both approaches A/B and MVP Testing can be used to test payment options, business models and complete products. It is critical, however, to set these tests up on an already pre-existing foundation of data. Before trying anything new, it makes sense to know how the web site is used currently.
Some companies are thoroughly immersed in analytics and apply the approach to everything they do. Zynga is one of these companies. Netflix another. In his book on analytics, Harvard Business School Professor Thomas Davenport describes the fascinating fact that an earlier phase of Netflix’s business model (based on shipping DVDs to customers instead of digital downloads) would not have been viable without meticulous analytics. The high shipping costs generated by active users of the service would have eroded the profits of Netflix. To retain their profitability, the company added a controversial “throttling” process which delaying DVD shipments for this subset of users. Netflix also uses detailed analytics and algorithms to recommend films to its customers. Finally, it extracts insight derived from viewer data of similar films as a negotiation advantage vis-à-vis media rights owners (source: Thomas H. Davenport and Jeanne G. Harris, Competing on Analytics, Harvard Business School Press, Boston, 2007, page 4).
Let’s step back a little from the details of different measuring and testing approaches. The ability to experiment on the web is not automatically a given, it is an aspect engrained into the very architecture of the internet. As Barbara van Schewick has pointed out in her excellent book on internet architecture and innovation, the so-called “end-to-end” architecture of the internet is the reason why rapid experimentation can happen at all. In a “core centered network,” innovation requires substantial systemic change on several levels. The current design of the internet allows rampant innovation at the edges, the “ends,” without requiring significant investment. In fact, as van Schwick points out using many great examples: Innovation can happen in one’s spare time (eBay, Del.icio.us, Yahoo, Facebook), can be paid for by consulting projects on the side (37signals, Blogger) or by family and friends (Amazon.com) (source: Barbara van Schewick, Internet Architecture and Innovation, MIT Press, Cambridge, 2010, pages 204 – 214).
While “end-to-end” architecture was part of the original design of the internet, software costs have fallen considerably year over year, due to the prevalence of Open Source and ready-to-go web services. In 1997, Silicon Valley-based venture financed internet companies still required hundreds of thousands of dollars to get started – this has been reduced to tens of thousands. Y Combinator, for example, provides companies with a comfortable starting package of $150.000, courtesy of two angels, Ron Conway and Yuri Milner.
As a result of falling costs, there has been a “remarkable increase in the degree of entrepreneurial experimentation,” according to Bill Sahlman of Harvard Business School. The same article that quotes Sahlmann also cites Apax founder Alan Patricof and his appetite for entrepreneurs that can “pivot” (source: „The pivotal moment. Bet on a boss who can twirl on his toes,“ The Economist, 04.12.10, page 74). In the history of its existence, Skype has changed payment options several times including trying out several variants of service and subscription fees. Twitter is perfecting its monetization scheme based on Tweet promotions continually as we are writing these words.
In an interview for the book Simply Seven, Frédéric Court, General Partner of the London-based fund Advent Venture Partners emphasizes the advantages of experimentation, adding that globalization of the internet provides further possibilities: “It is becoming easier and easier to launch a product extremely fast in different markets all over the world and see where it sticks. The whole world has become a business laboratory.” Court notes: “We love the globalization of opportunities via the web, especially as it enables start-ups in markets like Europe or the U.S. to remotely address high-growth emerging markets very efficiently.”
The costs of launching new ideas globally will fall even further. “Software as a service” (SaaS) and “cloud” offerings allow clever combinations of different services provided by different companies. This means that business building blocks will also be combined in new ways, too. Leading internet companies such as Apple and Google have started to offer payment options as part of their platforms, essentially extending web services into the area of monetization. We will discuss this more in the final chapter. As costs fall further, it will become even easier to experiment – these are exciting times.