This is the draft that I wrote for IEEE SPM's education column, to be published in April 2009.
Xuedong D. Huang
For many of us who have worked on technologies, a dream simply won’t quit. This dream is to bring the miracle of technology we have researched and refined for many years to the mass market. This dream is to offer the public a new feature, a new product, or a new service, derived from the great technology we helped to create, to make the world a much better place.
In 1993, I left Carnegie Mellon University to found Microsoft’s speech technology group with the dream that I could help to bring speech recognition technologies to the public. I have been fortunate to have the opportunity to work in both Microsoft’s research labs and business division to lead Microsoft’s speech efforts that have been successfully integrated into Microsoft Windows, Office, and Exchange. More recently, I am responsible for leading Microsoft’s incubation initiatives to create new communications products and services, one of such new products is Response Point (http://www.microsoft.com/responsepoint), an advanced phone system for the under-served small business customer. While it is too early to say this product will be successful in the marketplace, Response Point’s simplicity and iconic “magic” blue button with powerful speech recognition are loved by our customers and have received many awards and excellent reviews.
Figure 1. A Response Point phone has an iconic RP blue button (pictured above), which offers context-sensitive speech recognition.
In an enterprise environment, there are 3 major ways to start something new:
· Transfer technology from a research lab to an established product group, typically delivered as a concept, process, or a feature that can be integrated into existing products. Malvar has a good article discussing on this subject .
· Work directly in an established product group that improves technologies and services by releasing newer versions of existing products.
· Initiate a startup group and a new product to avoid many of the challenges that established groups and existing products face, which is the focus of this article.
Entrepreneurial efforts in established companies require dedicated people who want to change the world and affect the future. For brave souls in established companies, there are unique challenges to bring new products and services to market. I have learned many lessons in how to bring new products and services to market, while considering both the research and business points-of-view. Running a successful startup is not a science but an art. It’s my goal here to share the art of creating a startup. I believe these lessons apply to any endeavor, no matter if you’re starting a new product, a new company, or a new anything.
The fundamentals of how to start something new and compete in the market place haven't changed for decades. Like writing a research grant proposal, the first question to ask is if one can answer the basics for the business proposal. For a number of incubation ideas, I have used the following simple questions (why, what, and how) to evaluate and vet feasible and compelling proposals:
· Why. This explains the context and the fundamental reason for one’s startup proposal. Why should anyone care about the proposal? This could be a new technology based feature, a new way of using the existing technology, or a new architecture of using existing services.
· What. The key task is to describe the way things are and what unique aspects of the proposal could change the way things are. This further explains what one wants to achieve as well as what one doesn’t want to cover. Having a clear focus is critical to any endeavor in the early stage of the project.
· How. This explains what needs to be done to close the gap, including key questions on raise capital, recruiting, engineering, branding, marketing, and selling.
Most startups fail. Starting a new business in established companies is not necessarily easier than in a new company. Innovator’s dilemma is a universal challenge, as illustrated in . Kim and Mauborgne  outlined some very compelling ways of identifying a “Blue Ocean Strategy,” which is the result of a decade-long study of many strategic moves spanning more than 30 industries. It is the simultaneous pursuit of differentiation and low cost. The aim is not to out-perform the competition in the existing industry, but to create new market space, or a “blue ocean,” thereby making the competition irrelevant.
For Response Point, the small business phone market is very fragmented without having a clear leader - a classic example of “red ocean” landscape. We identified product complexity and affordability as the two key barriers that prevented 70% of small businesses from adopting an advanced phone system. As a result, Response Point focused on the product simplicity and competitive pricing simultaneously without worrying about a detailed feature comparison with incumbent phone systems, which became a key principle in guiding the team to bring the product to market.
Understanding the way things are in the real world is the first step toward developing improvements for the way things could be, which becomes the vision that drives the new initiative. Equally important is the execution that requires dedication and detailed understanding on what needs to happen to make the business successful. Many people often neglect to do their basics or keep the business plan current. A quick basic analysis can save a lot of wasted effort and help focus on the right strategy. Starting something new is hard. The ability to execute is as important as the ability to have a great idea. Most companies fail not because of lack of vision or strategy, but because of their weak execution.
As a company, Microsoft has always been trying to reinvent itself. The internal “startup” has been around since the inception of the company. As in many established companies, internal startups can leverage a number of key benefits, including great financial resources, established brands, intellectual properties, and other working infrastructure in place. Despite these benefits, creating a new product or service inside Microsoft is not necessarily without challenges.
Jack Welch, the former Chief Executive Officer of General Electric, said running an incubation of reaching $1m in GE is harder than running an existing business of $1 billion revenue. Nathan Rosenberg , a historian of technology, said “innovation is more likely to occur in a society that is open to the formation of new enterprises than in a society that relies on its existing organizations for innovation,” and “innovation often originates outside existing organizations, in part because successful organizations acquire a commitment to the status quo and a resistance to ideas that might change it.”
To address the innovator’s dilemma, one should first find a key executive sponsor to support the project. This is critical to get the project off the ground. For many of our startup efforts, we were able to have strong executive sponsors (Bill Gates or Craig Mundie in the case of Response Point) who provided advice, technical and business insights, and most importantly – protection.
Unlike running a startup from outside, in which one wants to get as much attention as possible to raise the awareness, it is a good idea to stay under the radar until either the project has enough progress or the rest of the company realizes the project is needed. Being small has a big advantage. Speed is associated with mass, as Newton’s laws taught us. In addition to speed, it is relatively easier to instill a sense of mission to rally the team to achieve a higher goal. There is no question a smaller team can accomplish the same kind of task that may take a larger organization more time required to ship the same product.
If possible, a separate building or group is preferred so the new venture can be left alone by antibodies in the corporate immune system. Many of our early projects started in Microsoft Research, which enabled us to keep the project under the radar and to take huge risks in the early stage. We are able to not only deeply leverage core technologies, but also address new opportunities that are across different Microsoft business groups. That very reason also made it a little bit harder for us to integrate the startup initiatives directly into one of the existing business groups, although the success of the startup should generally lead to integration.
Timing for “graduation” into the business group is important, which can be based on several factors such as if the new product itself is profitable financially or the larger organization has a strong desire to acquire the startup for the betterment of the company.
The startup’s mandate may be to create the new product or service that would put an end to existing products, or to focus on the cross-group opportunities that are neglected by the existing product groups. It is better for the internal startup to put an end to the existing products rather than have a competitor do the same.
Successful companies or products are usually created by a small team of good people. Collins illustrated this by studying a large number of successful companies . For any entrepreneur to be successful, it is important to have a good strategy, vision, capital, and technology, but it is most important to have the right team of people.
Working at Microsoft’s research labs and product groups for more than 16 years, I have been fortunate to have the opportunity to work with many different types of people. Most of the people I worked with could be characterized by the following features:
High tolerance for ambiguity
Low tolerance for ambiguity
If the project has too many people with personalities from the left column, it is very unlikely one will be able to excel in execution. I have seen many research results that have stayed in the lab become obsolete over time. If the project has too many people with personalities from the right column, it is possible that the team is working on a project that lacks innovation. Any product without differentiation won’t be unable to compete effectively in the market place.
These two types of people require slightly different ways of motivation and it is very important to build the trust among these two groups of people. It is a prerequisite for the startup leader to assemble a team that has the balance of both types of people. This is probably the most important success factor in retrospect for many new things I helped to start at Microsoft.
If brainstorming is the mother of innovation, I would think that benchmarking is the father of execution. Innovation and execution determine the success of running a venture.
It is necessary to challenge every assumption when one starts something new. The best way to come up with something new is to brainstorm with different kinds of people. While one makes progress, it is a good practice to conceptualize the vision via rapid prototyping and to verify the concept through benchmarking. One can have a mark-up demo or create a storyboard to encourage everyone in the team to think out of the box. Brainstorming is always part of my one-on-one meetings when I have an opportunity to engage with others. Brainstorming is always part of my learning experience. I can recall many innovative ideas were derived from engaging brainstorming.
For the speech recognition community, the golden period of rapid technology progress I personally witnessed was when DARPA institutionalized a rigorous benchmarking program for evaluating different speech recognition systems on a common task .
When one works on something new, it is guaranteed that it will be controversial. Otherwise, the task may not be innovative enough. The best way to silence different voices is to have data speak for the cause. This is equally important to convince and rally people in the team to move towards the same direction.
There is no fixed way of developing a new product. There are pros and cons with different approaches. Diversity helps to breed new innovations. I believe different approaches could all be successful with the right kind of leadership. Great products and ideas can come from anywhere and from any approach.
Microsoft has a set of time-proven software development processes that worked well for a large number of successful products including Windows and Office. For these established products and services, the engineering task is typically divided into development, test, and program management. A different business team is responsible for product management, business development and sales.
Developers write the code. They are in charge of the code's architecture, performance, reliability, security, and functionality. They are at the core of what we do as a software company. Testers insure the product ultimately meets the needs of the customer. They review all specifications and use their perspective and experience to ensure that the product is robust with a successful implementation. The role of program managers is unique. Developers are typically stretched too thin without focusing enough attention on the usability or the scenarios for how people would use the software. Microsoft created a specific role for program managers with the explicit goal of partnering with the development test teams. They create the specification for the product and work through the entire product cycle as the advocate for customers. In a typical established product group, the ratio among program managers, developers, and testers is typically 1:2:2.
In the early stage of the startup, we didn’t use the traditional Microsoft engineering process. Most people are researchers and developers. When we transitioned from incubation to the stage to bring the concept to market, we adopted a more traditional Microsoft process except that we put the engineering people and business people in the same team. The combined engineering and business team enabled us to have a more agile process to incorporate our learning from both business and product development perspectives. We focused on an “act fast, ponder less, and learn” philosophy and created many different prototypes and business and selling models to learn what really works in the market place.
We also adopted venture capital funding processes, which means we had very limited resources initially. This is effective as we really didn’t know what we wanted to do initially (like most startups). We spent a large amount of time listening to our customers, to understand their needs, to prioritize what we need to work on, and to set a realistic goal in the short term. This helped the team really focus on delivering “value innovations” for our customers and helped to create a market-focused organization. This staged approach also enabled us to take an end-to-end view of the business early on, which allowed for maximum flexibility and gave us clear accountability for delivering on the vision we put forth. After we obtained additional funding to bring Response Point to market, we found that Microsoft’s traditional engineering process, while it may not be as agile, was quite effective in delivering quality products and services.
The key learning is that we must adopt an adaptive process in the life cycle of the startup. We need to apply a different process transitioning from the early stage to scaling the business.
.... [read the whole paper @ IEEE Signal Processing Magazine]