IT Success!: Towards a New Model for Information Technology

5.5. Is prototyping the answer to everything?

Inevitably, when faced with two radically different approaches to doing something, each with its proponents, rational discussion on pros and cons usually gives way to evangelism and defensive posturing. Waterfall vs prototyping is no exception.

Which invites the natural question, 'Is prototyping the answer to everything?'.

Prototyping generally works well when performance and reliability are not critical on day one (ultimately performance and reliability are always important over time), and when 'good enough' today is better than 'perfect' tomorrow. Also, security and safety would not be key requirements, since by definition these cannot be simply 'good enough'. A typical example would be internal (i.e. non-customer-facing) CRM applications for sales and marketing.

Not surprisingly, prototyping generally works less well when performance, reliability, security and safety are critical right from day one. Such systems are necessarily more quality-driven than date-driven. Typical examples would be customer-facing systems, and applications for production and manufacturing. But even then, this needs to be qualified. It doesn't mean that prototyping is not applicable: it could still be used, for example as a first phase prior to using the traditional waterfall method.

A key factor in favour of prototyping concerns the dominant nature of applications being developed (or packages being bought) today. The first wave of 'back office' systems, ...