New Venture Development IT students and systems analysts traditionally have been brought up on the “waterfall method” in systems development. This formalises and codifies a set of stages from feasibility study through systems analysis, systems design, coding, testing, implementation to post implementation audit. It is logical, provides a structure and suggests review points for project planning and control.
However in the e-commerce arena, it has been found wanting. “It’s too linear”, “it doesn’t appeal to creative designers”, “it takes too long” and “it can’t cope with continuous ideas and change” were sample criticisms. Instead, systems development for e-commerce has become much more like the process of new venture development. Such an approach is most evident in web application development boutiques and dotcom start-ups where typically a system being developed is also a business development exercise – or a case of new product development. Indeed as in new product development, it is not unusual for some of the different tasks of the waterfall method to be executed in parallel in order to accelerate time to market.
However two other inter-related behaviours caught our attention and led us to explicate this venture development approach and propose a new model. First, aggressive time-boxing is the norm. For example, some companies will only undertake systems development projects that can be completed in three months or less. Many look to reducing the elapsed time of any stage to one month or less. Even redeveloping an early but fragile version of a system to industrial strength standard – a common need – mostly must not take more than nine months in total.
Second, the waterfall stages are being collapsed into three phases which, especially in start-ups, are very similar to the early financing phases of new business ventures: preseed, seed, and first stage financing. In the most illustrative examples, a team spends one or two days (but this can be one week to one month) on the concept, or preseed, phase. This involves brainstorming, storyboarding and rapid prototyping to develop a business or product idea and visualise it as a simple system. The team (see later) typically comprises business strategists, commercial managers, software developers, graphics and navigator designers, modellers and storyboard facilitators – and all may contribute to development of the concept.
The next design or seed phase typically lasts from one week to one month and produces a richer and more robust design. Graphics and navigation designers take centre stage here as do modellers and software developers. Usability and customer interaction tests are an important activity. The third development or substantial financing stage is dominated by the technology specialist whose job is to make the application robust. They may throw away the prototype from phase two ideal model of this “methodology” is proposed in Fig.2, where clearly the end of each stage can provide a review point, or in venture capital parlance, a financing decision. We borrow from the organisational development literature to propose labels for each phase of storming, forming and norming.
Programme Management Project planning and control is an essential management tool in traditional IT departments. The waterfall method provides a template and various mixes of time, cost and resource schedules are drawn up and then monitored. Change control procedures are commonplace as are devices whereby managements sign off on each stage. Often best practice or proprietary project planning and control systems are deployed.
In e-commerce ventures, timeboxing and deadlines are the major mechanisms used to plan and control systems development, in the spirit of time as currency. Typically, site or product launch dates drive the project and as deadlines approach, functionality (and sometimes robustness) are sacrificed or compromised in favour of time. “Delivery means delivery” was one observation we heard, because e-commerce projects are business projects with implicit or explicit promises made to the marketplace.
So project management has been simplified and project managers often do not exist. They are not affordable in start-ups and young spin-offs and as described in the process of rolling plans, progress and priorities are continuously reviewed in most e-commerce ventures.
However, a common learning has occurred in many of the firms we studied. Because systems are the product, the channel or the business process, there are other related activities that have to be planned, monitored and implemented in parallel. Examples include promotion, partnership agreements, legal matters and content acquisition. Thus start-ups, spin-offs and incumbents are experimenting with programme management – an office or manager responsible for overseeing and co-ordinating all systems projects, resource planning and tasks that have to be executed in parallel.
Launch and Learn The engineering ethos of traditional IT both grew out of and cultivated a “build to last” philosophy of systems development. As a result, there are businesses today running on core applications that are twenty-five years old or more. The term “legacy system” is certainly appropriate, but it is as much a complimentary term as a pejorative one. These systems have proved reliable and long-lasting. By comparison, the systems built for e-commerce often are disposable or throw-away applications. They are built on a “launch and learn” principle, driven by time to market goals, sometimes at minimum feasible cost and often to learn about the product, channel or market by doing.
One start-up described the systems activity as “doing, adapting and then doing it again”. One large incumbent reported that they “plan to throw away the first version”. A spin-off “will go to market even if the system is incomplete and linked to the back end with wires and tape”. An experienced spin-off finds the working life of a system so short – perhaps eighteen months to two years – that “it is not worth documenting either the system architecture or the coding”. An incumbent has learnt to abandon early versions of a web-based application when the business idea is clearly wrong, but never throws away the code because of re-use potential and the associated speed to market benefits that might be realised.
Launch and learn may be a transient policy appropriate when business models are evolving fast. It may even be a means of adopting the storming, forming, norming methodology and then reverting to the traditional waterfall method when requirements are certain. It is certainly one example of how the entrepreneurial style is being enacted. Multi-Disciplinary Teams The traditional IT department has been built around specialist jobs, each with their different status. Application programmers, systems programmers, systems analysts and project leaders have made up the typical hierarchy. Each role was clearly defined and demarcation between tasks was relatively unambiguous.
In e-commerce ventures, software developers, technology experts, graphics designers, business strategist, usability engineers, navigation designers, marketing specialists, modellers and media designers work collaboratively, especially in the storming and forming phases of the venturing lifecycle. While they bring their own experience and expertise, they often will contribute ideas or experiment outside their specialism. Demarcation lines are very thin and everyone or anyone can contribute to the system concept and design if it advances the endeavour in a constructive manner.