Feasibility and Desirability analysis Colin Potts Georgia Tech

Feasibility and Desirability analysis Colin Potts Georgia Tech © Colin Potts, 1996 C 3 -1

Requirements’ priorities l Not all requirements are equal » Even contractual requirements are often renegotiated » Product requirements are often assigned priorities – low priority reqts. may be jettisoned to reduce time-tomarket – Priorities may be {1, 3, 9} or {possible, deferred, rejected}, etc. l Priorities depend on stakeholders » Different stake value reqts. differently © Colin Potts, 1996 C 3 -2

Overview: QFD and prototyping l Quality Function Deployment (QFD) » Highly structured approach for tabulating reqts. /design interactions & evaluating alternatives l Prototyping » Early development of executable models for customer to evaluate system concept © Colin Potts, 1996 C 3 -3

Quality function deployment (QFD) Japanese technique for design & manufacturing l Principal tool is the “house” of quality l » Complex matrix showing dependencies & trade-offs » Simultaneous analysis of customer needs, engineering design & market data lends QFD to multi-function teams © Colin Potts, 1996 C 3 -4

Building a “house” Design attributes Customer needs (named reqts. ) Customer evaluations Eng. measures © Colin Potts, 1996 C 3 -5

Prototypes l Prototype construction is standard practice in engineering » cf. wind tunnels, computational models, artist's impression, etc l Design involves much exploration » Requirements can be validated by prototyping – To explore whether requirements really reflect users' needs » Prototypes reduce risk – Prototypes are used during feasibility studies » Prototypes can be constructed from reusable parts – Typically use 4 GL, interface builders, application libraries, etc. © Colin Potts, 1996 C 3 -6

Effectiveness of prototyping l Empirical studies » Boehm et al (Proc. ICSE, 1986) » Alavi et al (CACM, 1986) » Most evidence is anecdotal l Experiment: incremental development (P) vs. specification-based design (S) » » » Student team projects to build estimation package No difference in performance P led to 40% smaller code P required 55% effort of S P led to less functionality, poorer reliability and poorer maintainability, but was easier to use and learn © Colin Potts, 1996 C 3 -7

Effectiveness of prototypes (cont. ) l Analyst and manager survey » $5 K to $630 K projects » Perceived benefits of prototyping – Better feedback about requirements – Prototype provided common medium of communication (users and experience develop better working relationship) – Users become more enthusiastic » Perceived limitations of prototyping – Easy to oversell – Difficult to plan and control prototype development – Temptation not to throw away © Colin Potts, 1996 C 3 -8

Scenarios and storyboards © Colin Potts, 1996 C 3 -9

Feasibility & desirability: how to find out more l QFD » Articles in HBR etc. l Prototyping » Many articles in IEEE Software and elsewhere » Book by Vonk © Colin Potts, 1996 C 3 -10