Analytica 277

experience but no experience with software products. Taurus had to relocate to be with the six-member Analytica team that was building the company's future product. The head of engineering, Brad Silverberg, an engineer on Lisa, came from Apple with Eric. The CEO, two engineers, and three members of the support staff from Taurus remained with the company after a year.

Figure 11-2 shows the relational graphs for the Analytica start-up at the concept and seed stages. Notice that at the concept stage, the diagnostic is able to pinpoint the flawed business plan, given the newness of the product concept, the lack of specification, and poor market understanding. The lack of a CEO also shows up right at the concept stage.

By the end of the seed stage, when the venture capitalists created the merged company, the firm was already in serious trouble. Clearly, there was no vision and plan for the company as a whole, since Analytica was nothing like what the founders had originally proposed and the new president did not prepare a new plan for the combined organization. Whereas marketing was behind in planning, sales was ahead, in view of the fact that it was assembling a sales staff. The board at this point was OK, even though its members lacked software experience and did not act conservatively with respect to control. There was as yet no indication that the firm was out of control, except that the merger with a failing company, by definition, meant that the new organization was unable to meet its plans. The lack of a clear product specification at the seed stage was a sign that both the technology/engineering and the product had problems right from the start.

PRODUCT DEVELOPMENT STAGE

With the company in full swing, everyone had ideas for the product, including the fully staffed marketing and sales groups; the new, strong engineering team; and the board. The product specification for Reflex advanced the state of the art in the graphical user interface; in database design, including implicit data typing; and in the ability to add and subtract fields within the database. The product also had a variety of analytic powers, with the ability to do charting. For example, it could make a chart by extrapolating from a historical database.

However, for a product with such a plethora of features, Reflex had a weak underpinning because of the decision to run the program "in core" instead of working with a virtual memory environment. The fact that the program had to reside in memory at all times significantly limited the size of the applications and the underlying database. This was almost, by definition, a fatal flaw from the standpoint of potential corporate users (particularly since Analytica planned to market the product according to the Lotus model) and was therefore a misinterpretation of the market. After a year, Eric and Brad were finally granted explicit control over the product specification.

Given the program's size and complexity, the fact that it constantly grew in functionality, its lack of definition, its complex but undefined graphical user interface, and the naiveté' of everyone concerned with the schedule, it was inevitable that Reflex