The New York Times
Printer Friendly Format Sponsored By

February 8, 2007
Economic Scene

Kaizen, That Continuous Improvement Strategy, Finds Its Ideal Environment


Remember when Japanese manufacturing techniques were all the rage? You could hardly read the business press without encountering mention of “lean manufacturing,” “just-in-time inventory systems” and “total quality management.”

You don’t hear much about these ideas anymore, but not because they are no longer in fashion. Quite the reverse is true: the practices have become so widely adopted that they are no longer newsworthy.

Nowadays, it is American executives — especially those from high-tech industries — who are lecturing the rest of the world on the wonders of Web 2.0 and the latest new new thing emanating from Silicon Valley.

It is therefore a paradox that one of the most important drivers of online business success is taken directly from the pages of Japanese management techniques. I am referring to kaizen, the practice of continuous improvement.

Kaizen doesn’t just mean a business should keep trying new things. Rather, it refers to a disciplined process of systematic exploration, controlled experimentation and then painstaking adoption of the new procedures. In the original formulation, kaizen was applied to manufacturing, where experimentation could determine whether a new process resulted in quality improvements or cost savings in a matter of months.

It is much more difficult to apply kaizen to product design, since it can easily take years to design and market a new product. To take a recent example, the iPhone has been two and a half years in the making.

Product development can cost hundreds of millions of dollars, making it almost impossible to run a controlled experiment with a product introduction.

But it is simple to run a controlled experiment with a Web page. Amazon can show a different page layout to every hundredth visitor and determine in a few days whether the new design increases sales.

Similarly, a search engine can run a controlled experiment to try out a new tweak to its search algorithm and discover in a few hours whether users find it an improvement on the old algorithm. On the Web, continuous improvement really is continuous.

The cycle of exploration, experimentation and adoption is drastically shortened for Web-based applications. This isn’t just the old atoms and bits distinction. Vista, Microsoft’s new operating system, has also taken years to develop and only time will tell how successful it will be.

What’s the difference between Vista and Google? There is no feasible way for Microsoft to experiment with Vista in real time; but it is very easy for Google to conduct controlled experiments and do so more or less continuously.

Given a performance measure, be it clicks, revenue or something entirely different, a disciplined process of experimentation and evaluation can lead to rapid improvements. The easier it is to experiment and the larger the number of users, the quicker this process can work.

The most successful online businesses are built on kaizen, though few of those who carry out the testing would recognize the term, since many of those who created these online businesses were in grade school in the 1980s.

Old media just do not understand online kaizen. Their perceptions are tied to the print world, where design changes are costly. The Wall Street Journal spent years planning its recent redesign of the print edition and millions of dollars rolling it out. Yet it will be months before it becomes clear how successful these changes were.

By contrast, small tweaks in the page layout of online content can be very effective in improving user satisfaction and ad clicks. Controlled experiments can be used to determine the impact of these changes in days rather than months.

Yet how many mainstream publishers have Web page software that allows for such controlled experimentation? In most cases, there is but one layout, and experimentation is difficult if not impossible. You can’t manage what you can’t measure — and if you can’t easily experiment with what you are doing, management is seriously handicapped.

Kaizen means that the companies currently in an industry have an inherent advantage over new entrants. Entrants have to guess what will work; the companies that are already operating can experiment and find out.

This information advantage doesn’t preclude new entries; it just makes it more costly since the learning curve is steeper.

Amazon has some worthy competition in online department stores. But how likely is it that a new entrant will emerge from nowhere and successfully compete in this area? The experience that existing online retailers like Amazon, and eBay have built up is hard to duplicate. A new entrant, even one as strong as Wal-Mart, finds the online world rough going.

This is not to say that new entry is impossible. As the old saying goes, “You can always tell the pioneers, they’re the ones with the arrows in their backs.”

New entrants have the advantage of avoiding earlier mistakes. They can copy successful operations and, in many cases, improve on them. Newer, faster and more flexible information systems can sometimes confer a competitive advantage on new entrants over the pioneers stuck with the last generation of computing infrastructure.

But the ability to experiment easily is a critical factor for Web-based applications. The online world is never static. There is a constant flow of new users, new products and new technologies. Being able to figure out quickly what works and what doesn’t can mean the difference between survival and extinction.

Hal R. Varian is a professor of business, economics and information management at the University of California, Berkeley.