HE ONE ue Zv O Pae62 10th in a serteš on rapid appli- cations develop- ment (RAD). Joint applica- tion design (JAD), originally introduced by IBM Canada, greatly facilitates "the communica- tion of specifica- tions from end users to informa- tion systems (IS) analysts. The technigue, particularly successful for applications development, has spread to many corporations. It speeds reguirements analysis and de- sign, and yields designs that, better meet end users' needs. The basic idea of JAD is to select, key end users and conduct workshops that progress through a planned set of steps designed to create the system rapidly. In the session, IS staff members translate users reguirements into relevant data models, screen and report designs, pro- cess-flow diagrams, decomposition dia- grams and rough prototypes. Early JAD sessions were done without automation, but the process is far more effective when combined with integrat- ed computer-aided software engineering (I-GASE) tools, which can generate code for complete prototype applications. The skills of the person who organizes and conducts the JAD workshops are particularly critical to their success. Usually one person, called the JAD lead- er, runs the session. The job of JAD leader should be re- garded as a profession, reguiring profes- sional skills that take time to develop. JAD leaders are not likely to conduct their first session perfectly. Only after three or four sessions does the leader gain the necessary confidence and skill to make the session as effective as possi- ble. When a JAD session fails, it is al- most always the fault of the leader. JAD leaders should be chosen mainly for their communications skills. They may come from the end-user or IS com- munity, but in practice, most come from IS. Some JAD leaders have been infor- mation-center staff members, some have a marketing background, and some are The JAD leader needs to be diplomat- ic and disassociated from any politics that might affect the session. Above all, the leader must remain objective. It is his or her job to prepare the session, orchestrate the interaction between the participants, confine discussions to a structured framework and move the session relatively guickly to the re- guired conelusions. The JAD leader acts as the focal point for the views of management, end users and IS profes- sionals. The JAD leader needs to research and prepare for the meeting, and to provide the participants with appropriate print: ed material. Additionally, JAD leaders should be: ' comfortable working. in front of a group of people, j m. m. A o a some ——— mm O Ee APP INTELLIGENCE sovi he Poasvep 1 Success of JAD Workshops Depends; L. 8 confident in their ability to Pia dis- cussion and fact, finding; PO MEEKNAPPLIGATION DEVELOPMENT € able to command the respect of all parties at the session; € well prepared, zih € knowledgeable about the area of business; 8 competent in the technigues being . used; and s able to control controversies and stay flexible. A good JAD leader knows that cer- tain goals must be accomplished by a JAD leader moves the i prem zma E, pe zes ': PRE a ei EH — | E vame So — a ki | H eparaled into subsystems with an ]- 4 CASE tool that ensures consistenit data - models and precise interfaces between . the subsystems. Separate, one-week JAD v sessions may then be used for the sepa- 'rate subsystems. | j Participants in the JAD workshop 4 should understand four types of dia- ' grams: entity-relationship diagrams; de- ' composition diagrams; dependency, or data-flow, diagrams; and action dia- ' grams (used to specify procedural logic). > These can usually be taught to end us- | ers in a half-day course. In some organi- given time. The session forward until the reguisite de- signs are Sirene along with screens, reports and, pesi prototypes. The sessions are used. ' zations, however, two half-day training Users who have not previously partic- Essentlal Eleriients of a New |.S. Professional: Joint Application Design Leader tity to blend disparate View Ži Va a alk di E Protessiorials dij to direct gob and comma "Spect of participant be, JAD ih Strong leadershiP skil? | L i V b r "1 A " ozi negi m "a rabe ba > kim Avakin sj ne bi | | dopdi ei ! EME ž ie 4 O ALI O delay. It is important for the JAD leader to be enthusiastic. a ee ae | goal is to discuss ideas fully and reach agreement without much delay. It is im- portant for the session leader to be en- thusiastic about, JAD and to convey ex- citement to the participants about how: well it can work. A corporation may employ an butaddo. consulting firm to run its first JAD workshops until its own JAD leader be- comes experienced, For RAD, | recommend full-time—not half-day—participation in JAD work- shops. Sessions should usually last five 44 ] iu LI 1 a, i ui 4 ! ubi z) ij Re aijE | dpated in JAD need to become familiar (with what it is and what they are ex- f ega to contribute. The users and the JAD ade: SEE h ole k. ta U ani discuss the diagrams on a white, i | board or flip charts. A seribe enters the , ( diagrams into an I-CASE tool set, which - shows how the diagrams relate to one another and provides details of the ob- r jects. The participants examine the rela- - Honships and details on a large-scree | , moni tor, adjusting the computerized de- Len gosto Horkahop Pp Dgresses days, since participants tend to spend . NE a RAD EN ia Vere EN tvo the first two days getting to know one JAD workshops in tne user design another. a | ' phase, Before the first workshop, there Very large applications need to be. vede a period of preparation, and 8 kickolI ne a ai a a a ei a a arge - $ building an initial data-flow diagram H more detail; « shown screen designs and prototypes of ' the dialogue. They should examine and :, comment: on reports tni , produces. ; 'solidified and cleaned up by IS profes- ' sionals, and prototypes are built; users ' reviewed, along with the output of the | paje joint reguirements planning FF EBRUARY 26, 1990 Iy on Leaders meeting is held. The initial design of the system is done in the first, workshop. To begin, the scope and objectives are P) workshop. Relevant information from the I-CASE repository is reviewed. It may be shown "live" on the work- shop's large-screen monitor using the [- CASE tool. Adjustments may be made and comments added. The data model may be discussed and modified. | Each process, represented by a pro- cess block on the process- flow PASTI is then examined in detail. This may be 4 done in five stages: $ determining the steps in the pro- cedure; showing the steps; 6 examining each procedural step in $ creating a partial prototype for each procedural step; and € addressing unresolved issues. The workshop participants should be h'system After the first volkom: the design is examine the design documents generat- "ed by the I-CASE tool and work with — the prototypes. At the second workshop, the experi- -ence with the prototypes is reviewed and enhancements to the design dis- . cussed. Substantial design improvements . phase should overlap the JAD activity, ' cess. The construction team becomes in- — comfortable with the design to give the "workshop. : which may indicate that. the system be : parallel by separate teams. — Next week, | will discuss the concept The concepts embodied in RAD are de- scrwbed tn a new volume in the James 1240. For information on semtnars, or additions may be made. After the second workshop, the design is further -solidified and then finalized. Preparation for the construction taking place near the end of the pro- volved after the first JAD workshop, when the design is being consolidated and the prototypes evaluated. Members of the team are present at the second The construction team should esti- mate the amount, of work reguired, split into subsystems to be developed in The goal of the second workshop is to make a presentation to the executive owner and allow him or her to guestion the participants. The owner must feel go-ahead for construction. of SWAT teams—small, highly trained teams of IS professionals who build the application under deadline. M Martin Report Series. For more infor- mation on this volume, call (800) 242. otok (is (ke United States and Can nology Transfer Institute, 741 10th aaa Monica, Calif. 90402 (213) 394-8305. In Europe, con- tact Savant, 2 New St., Carnforth, LA5 9BX United Kingdom ( (0520) 734 505.