The Agile Revolution
Innovative Product Development
Product development teams are facing a quiet revolution in which both engineers and managers are struggling to adjust. In industry after industrypharmaceuticals, software, automobiles, integrated circuitscustomer demands for continuous innovation and the plunging cost of experimentation are signaling a massive switch from anticipatory to adaptive styles of development. This switch plays havoc with engineers, project managers, and executives who are still operating with anticipatory, prescriptive mindsets and processes geared to a rapidly disappearing era.
Symyx creates and operates highly integrated, complete workflows that enable scientists to explore their ideas to discover and optimize new materials hundreds to thousands times faster than traditional research methods. These workflows consist of robotics that synthesize arrays of materials on a miniaturized scale, creating hundreds to thousands of tiny experiments on one silicon chip. These materials are then rapidly screened in parallel for desired physical and functional properties, including chemical, thermal, optical, electronic, or mechanical attributes. The results are captured in database systems for mining large sets of data to help scientists make well-informed decisions on the next steps of the discovery process.1
Symyx boasts 100 times the speed at 1% of the cost of traditional research. Drug companies used to design compounds for specific purposes by having scientists pore over how to make just the right one. Today they generate tens of thousands of compounds and then test them quickly using ultra-sophisticated, ultra-speedy tools such as mass spectrometers. There are new product development economics at work here.
In mid-2002, when Alias Systems of Toronto, Canada, started developing Alias Sketchbook Pro, a software package to be announced concurrently with Microsoft's launch of its Tablet PC operating system, the product management and software development team didn't begin with a lengthy product planning effort. The team's marketing and product strategy work evolved over several months, but its product development effort began early, and in parallel, with the strategy process. The team had a visionan easy-to-use consumer-focused sketching product worthy of a professional graphics artistand a deadline, the November Microsoft launch date. The product evolved in two-week iterations. For each iteration, a short planning session identified features to be developed. Then, within the "platform" architecture constraints of the operating system and Tablet PC computers, the product evolvediteration by iteration. In the end, the product was delivered on time, met high quality standards, and has been a success in the marketplace. The product wasn't planned and built, it was envisioned and evolved. Alias didn't start with anticipated architectures, plans, and detailed specificationsit began with a vision followed shortly by the first iteration of the product. The product, the architecture, the plans, and the specifications evolved as the team adapted to the ever-unfolding reality of the market and the technology.
With Alias Sketchbook Pro, the team literally didn't know past the next iteration which features would be included in subsequent development iterations. Team members did have a clear product vision and a business plan. They did have a general idea about what features were needed in the product. They did have active involvement from product marketing. They did have an absolute time deadline and resource expenditure constraints. They did have an overall product platform architecture. Within this vision, business and technical constraints, and overall product release plan, they delivered tested features every two weeks and then adapted their plans to the reality of actual product testing. The team's process was one of evolution and adaptation, not planning and optimization.
In the automobile industry, Toyota employs set-based design in its design processmaintaining multiple design options on components until late in the development process. Similarly, BMW uses simulations to improve car crashworthiness. During one design effort, it ran 91 simulations and two real crashes. The results were a 30% improvement in design and 2.5 days per simulated crash versus 3.8 months (for simple tests)and the 91 simulations cost less than the two real crashes (Thomke 2003).
All of these approaches to product development point to a very critical issue. When we reduce the cost of experimentation enough, the entire economics of how we do product development changesit switches from a process based on anticipation (define, design, and build) to one based on adaptation (envision, explore, and adapt). When the cost of generating alternatives plunges and the cost of integrating them into a product is low, then great products aren't built, they evolvejust like biological evolution, only much, much faster than in nature. Biological evolution begins with experimentation (mutation and recombination), exploration (survival of the fittest), and refinement (producing more of the survivors). Increasingly, product development processes are being built using this analogy.
Time is also a driving factor in new product development (NPD). In the short and intense decade of the 1990s, the average new product time to market in the US dropped from 35.5 to 11 months (Wujec and Muscat 2002). "Corporations everywhere are engaged in a new products war," says NPD expert and author Robert Cooper. "From soup to nuts, from can openers to automobiles, companies are at a competitive war with each otherand the advance troops are product development teams. On this new product battlefield, the ability to mount lightning attackswell-planned but swift strikesis increasingly key to success.. . . And mobility or speed enables lightning strikes designed to seize windows of opportunity or to catch an enemy off guard" (Cooper 2001).
But uncertainty, shrinking time schedules, and the need for iterative exploration are not restricted to new product development. New business practice implementations, such as those fostered by customer relationship management (CRM) installations, are often fraught with uncertainty of a different kind. The high rate of failures reported in CRM implementations can, in part, be attributed to anticipatory (plan-driven) project management approaches that failed to "explore" into the uncertainty caused by major business process changes. Companies tried to plan and do when they needed to envision and explore. As authors Preston Smith and Guy Merritt (2002) write, "Innovative product development depends on exploring the uncertain to add product value and maintain competitive advantage."
But innovation and faster development aren't good enough. Companies have to deliver better products geared to what customers want at the time of shipment, which may or may not resemble what the team guessed they wanted when the project was initiated. Ultimate customer value is delivered at the point-of-sale, not the point-of-plan. Companies that have the ability to quickly and inexpensively evolve a product closest to the end of the development lifecycle will have a tremendous competitive advantage.
So why isn't every company doing this? Because for most companies there is a great gap between needing and delivering new products. NPD is a multifaceted and extremely difficult challenge. The Product Development and Management Association (PDMA) estimates newly launched product failure rates of around 59%, which has changed little from the early 1990s. Also, cancelled or failed (in the market) products consumed an estimated 46% of product development resources (Cooper 2001). Yet some companies are consistently more successful than others, and a growing number of these successful companies are practicing agile development and project management methods.
The product development efforts targeted by agile methods include new products2 and enhancements to products in the domains of:
Software products (examples are PC applications, operating systems, middleware products, enterprise resource planning)
Industrial products with embedded software (from electronics equipment to autos)
Internally developed IT products that fit the speed, mobility, and exploration factor criteria
The key point is that opportunity, uncertainty, and risk reside in the proposed productnot in the approach to project management. Our approach to project management needs to fit with the characteristics of the product in order to improve our chances of capitalizing on the opportunity by systematically reducing the uncertainty and mitigating the risks over the life of the project.
Companies need results from their high-pressure product development efforts, but they shouldn't come at the expense of quality. John Wooden, the legendary basketball coach of UCLA whose teams won 10 national championships, used a saying with his teams that applies to product development: "Be quick, but don't hurry." In other words, do the right things, but learn how to do them quickly. Strip away the overhead, the non-value-adding activities. Create quality products and do it quickly. Agile development focuses on speed, mobility, and quality. To accomplish this, individuals and teams must be highly disciplinedbut with self-discipline rather than imposed discipline. Anyone who practices ad hoc development under the guise of agile methods is an imposter.
There is no reason to think the changes in the next ten years will be of less magnitude than those of the previous ten, although the emphasis will likely change from pure information technology to the integration of information and biotechnology. The underlying codes of information technology are zeros and ones. The underlying codes of biotechnology are A, T, C, and G (the components of DNA). When biological codes can be reduced to computer codes, as in the Human Genome Project, and then be manipulated by computer programs (as is happening), the potential impact on product development of many types is staggering. "New materials, programmed molecular factories, and self-organizing fabrication processes could change the cost and performance characteristics of everything from drugs to dragsters, paint to plastics, china to chairs" (Meyer and Davis 2003). Scientific and technological advances in the coming decade and beyond will continue to irrevocably alter product development processes, and those changes, in turn, will cause us to rethink the management of those processes.
Linear thinking, prescriptive processes, and standardized, unvarying practices are no match for today's volatile product development environment. So as product development processes swing from anticipatory to adaptive, project management must change also. It must be geared to mobility, experimentation, and speed. But first of all, it must be geared to business objectives.