Making a good platform

Why is a platform business model attractive

Success of Google and Facebook has certainly given credibility to the platform business model. A well designed platform can engage multiple groups of suppliers and buyers. And, as the engagement increases, it can build an ecosystem that is self-reinforcing which becomes its own competitive barrier.

The initial mental challenge with platforms is to NOT think exclusively of monetizing the final user base. Take Google as an example. On one hand, you want to make sure that your actual buyers, i.e. the advertisers, are getting good value. On the other, you want to ensure the end user experience remains relevant to keep the eye balls, which is what the advertisers are paying for in the first place.

Once you have figure out the balance, focusing on your key success factor actually becomes straight forward, albeit not easier. To borrow a well used business strategy jargon, you can really concentrate on your core competence. This is not to suggest that you abdicate the responsibility of engaging with your ecosystem. But, it does mean that you can more effectively leverage existing channels and tap into unmet demands.

A few observations on platforms

Prof. Kevin Boudreau of London Business School and Prof. Andrei Hagiu of Harvard Business School have written Platform Rules, an insightful paper on aspects of regulating a platform. This is an unique challenge since a well regulated ecosystem is fundamental in attracting participation.

Conversely, a fair question on platforms is how much control could a company exert in the functioning of the ecosystem. Specifically, how far can a company push its own product or agenda before the fear of lock-in surfaces?

Finally, platforms is not an exclusively software idea. Take Apple iPhone as an example. It is a very well designed product, like all Apple products. But, good design is an insufficient condition since PC is still the most dominant format. I would argue that one of the reasons that iPhone become such an iconic product is the fact that it is supported by a platform called iTune/App Store. As a platform, iPhone allows consumers of digital goods, movies and app's, to connect with suppliers.

In the example of iPhone, Apple gets paid by the end users when they buy an iPhone. App writers pay Apple's iTune market a fee to access the iPhone users. And, phone carriers provides the technical and product distribution network to proliferate iPhone.

Platforms and business model

Business Model

In the most simplistic definition, business model is about how a business goes about making money. Typically, the discussion is concerned with the ways in which information, resources, suppliers, and buyers are organized.

The traditional model is somewhat of a black box where information and resources are processed and suppliers and buyers are aligned for input and output, respectively. Traditional manufacturing companies such as GE would be a good example. Even in the hi-tech arena, this is the norm until recently. For example, Oracle hires programmers, upgrades its database and services offerings, then sells to users.

The benefit of the traditional model is that a company has decent control over its destiny since it controls the direction and resources. The drawback, however, is that there are significant competitive barriers when you want to move away from what a company has traditionally done into a new space or a new way of doing things.

Platforms

Business models for tech firms, most pronounced in software, have been shifting to the concept of platforms in recent years. Instead of seeing a linear value-chain, these companies compete on the ability to connect supply and demand, like a platform. Beyond the ability to foster transactions, what makes a platform robust and successful is the ability to create a virtuous cycle whereby participation by one group can increases participation by the other group(s).

Examples of Multi-Sided Platforms

Prof. Andrei Haigue of Harvard Business School has been exploring the notion of multi-sided platforms in great depth. In his Note on Multi-Sided Platforms (MSP), here are a few examples of MSP.

• E-commerce sites like Amazon mediates between buyers and sellers
• Search engines like Google mediates between online advertisers and users
• Social media sites like Facebook mediates between users, advertisers and third-party application developers

A practitioner's notes on real-options analysis

The process

Having led my share of real-options analyses, the key thing to keep in mind is to split the tree into the three key phases of a product - technical, commercial, and users. The key insight for the technical phase is whether the solution will behave as advertised. For commercial, you want to find out if a product will be accepted as part of the BU's portfolio. For the user phase, it is all about how where does it fit within the existing usage landscape.

This then provides a rough guide on what kind of information will be needed and where you can look for the best information. Within each phase, each branch is a discussion point with the relevant experts. And, what I find most revealing is to use simple arithmetic averages as the starting point, i.e. if there are two branches for a node, it would have a probability value of 50% each, then ask the expert to adjust the percentage and why the adjustment.

The critical junctures of the analysis are the transition points. The fact that a technology is superior than the alternative, think Betamax vs. VHS in the original video format war, is no guarantee that it will be the winner. For the technical-commercial transition, one of the key question that I like to find out is the current product roadmap. The roadmap will provide implicit guide on what kind of technical performance is expected and how a new product will fit into the BU's direction. Another good reality check is if the BU has a history of commercializing new ideas beyond incremental improvements. Adding a feature, however small, is always a shock to the system and you want to make sure that there is a good way to deal with corporate anti-bodies. For the commercial-user transition, I like to understand how users are currently dealing with the problem my solution is supposed to solve. Inertia is a very powerful thing and the fact that you have come up with a better search engine does not mean Google can be dethroned tomorrow.

The toughest type of real-options analysis is when the application(s) is truly novel; therefore, most of the future numbers are best guesses only. For example, although people have developed technique of making CMOS semiconductors on polymer substrate in applications ranging from Organic LED to flexible sensors for a good many years, very few of them have made a meaningful transition to the commercial phase as a part of a product roadmap. In short, this makes it really hard to pass the senior management's "gut check." On the other hand, using the real-options format, you probably have the best chance to convince management to consider on-going investment in this type of project because highlights the option and (minimum) costs to play in an emerging market.

Real-options as a communication tool

Real-options is really about communicating insights in a structured format. An often underestimated benefit of using real-options analysis is that it allows diverse opinions to co-exist. Unlike the traditional NPV (net present value) analysis where everyone is angling for the right discount rate, cash-flow, or terminal value, real-options allows discussants to break down the overall project into logical components so the pessimists can have their say as do optimists. And, as new information is discovered or assumptions confirmed, the tree structure allows the overall integrity of unrelated information be preserved.

A full blown real-options analysis can be a bit overwhelming for the senior team. On the other hand, an appropriately high level reduction is usually a very informative tool for discussion. Conceptually, it should be made intuitively clear that a Yes decision for an investment is really an option to review its progress at the next milestone and any projects will go through several milestones. As long you don't get caught up with the details, real-options is a good way to verbalize/visualize this process.

Finally, what is the appropriate level of reduction for high level management discussion? This is where you want to conduct sensitivity analysis of the structure and hone in on these. Make sure that these assumptions pass basic reality check.

Real options analysis

Turning uncertainty into an advantage

The fundamental challenge of innovation is uncertainty. Taking a long time is not a problem if you know exactly how much time is needed. High initial cost is less of a concern if you are confidence of how many units you can sell over its lifecycle. Annual planning is much more straight forward when there is no viable alternative that competition can use to bypass your innovation. These are all reasons why innovation gives many executives the hibby gibbies. As a matter of a fact, this kind of uncertainty often leads to a condition called "analysis paralysis" where there is always one more piece of data that can be added to reduce the uncertainty.

But, with a bit of planning, this uncertainty can be turned into a competitive advantage. Be you a VC, a corporate development head, or an innovation officer, what better way to outrun your peers with a systematic way of reducing the many uncertainties in innovation.

Horizon planning and tech vs. market matrix

Horizon planning is usually used for senior management planning. It is a way of segmenting the strategy by time. Horizon One is the most immediate with concrete products and customers where most of the attention is paid. Horizon two and three are further out in time and include a higher level of uncertainty as well as long term projects. The strength of horizon planning is that it allows the management to clearly articulate what needs immediate attention (Horizon one) while acknowledge and make time for longer term activities (Horizon two and three.)

Another tool that seems to be used regularly in the CTO's office is the technology vs. market matrix. While the exact flavor differs, technology goes from Core to New and the market goes from Existing to New. The resultant four quadrants allows you to plot where all projects are from legacy to game changing. If a project falls into Core+Existing, you can usually get the BU to pay for it. Conversely, if a project resides in the New+New quadrant without a strong sponsor, you would have a lot of explaining to do (better yet, don't even go there.)

Both Horizon planning and the two-by-two matrix are great ways to convey a lot of information to the busy senior team. The one common failure point, however, is that these only provide a snap shot of what is going on today with no consideration for even known uncertainties.

Real options analysis

Real options analysis took its cue from the world of financial options. Most projects, like financial options, have multiple variables that would impact its outcome. To the extent that financial options model, namely Black-Scholes, provides a reasonable way to approximate the value of an option at any point in time, given what is know; the aim of real-options is to do the same for real projects in the real world.

The basic insight is that you view the possible outcomes in a tree structure. Each key milestone, e.g. technical, commercial, or regulatory, has impact on the value of the project. So, unlike the more traditional analysis using NPV calculation where there is only one possible value per variable, a real-options analysis incorporates a range of possible values while giving a single NPV-like value at the end.

Many people have worked on this topic. Professor Ian MacMillan of Wharton at University of Pennsylvania have done extensive work on this and other issues related to the innovation process. For a practitioner perspective, Mr. Scott H. Mathews of Boeing has worked on this topic and came up with the Datar-Mathews model which provides a convenient way to model real-life projects.

Add this management tool

I like incorporating real-options in the management conversation, in addition to other tools like horizon planning and matrix. It is a good way to explicitly acknowledge the dynamic nature of the market place. At the same time, because it does derive a single value, the result is simple enough so that it is not going to scare senior team from tuning out the analysis.