I am currently involved deeply in a project that replaces an exisiting spreadsheet-based production-planning system. The existing user-maintained document is one of the most dangerous spreadsheets I have ever seen.
But when contacted to help on this I made a terrible mistake right up front by not applying one of my most iron-clad rules of development – namely:
“The results are ALWAYS better for the client, the Business Process and for myself when I do an application from scratch rather than trying to fix an existing model or database with chicken-wire and bandages…”
In this case we have several long-term employees who are very afraid of change and who rely on this document for all their planning and ordering for a series of manufacturing plants. And I caved…
Whether this should be done in a spreadsheet is a legitimate question, and considering I kind of agree with that, I have integrated data automatically downloaded from Corporate EDI systemsinto the backend and now a process that took 2 people the better part of a day to do manually (and with a HUGE potential for error) the population now takes about 25 seconds, eliminates errors in transposition and leaves them time to do the analysis and real planning they should be doing. But by integrating the old 800+ row model as the presentation layer the process has taken much more time and effort in design and rconciliation than if I’d stood my ground up front.
Another lesson re-learned ….