When a business fails, the principals involved often have different opinions about why it happened. When Codeworks failed, there were a lot of reasons to choose from.

Although the firm was relatively small -- with 45 employees at its peak -- three of its top executives have different interpretations of why this well-funded enterprise couldn't get off the ground.

Codeworks was formed in Washington in 1986 to create an electronic database of building codes and standards, the complicated regulations that govern such requirements as the length of hallways and the placement of sprinklers in commercial buildings.

What Codeworks did that hadn't been done before was to add local amendments to standard regional building codes, so architects and engineers could order reports tailored to a specific kind of building in a specific city or county.

If it had worked, it could have changed drastically the nature of building code research, much the way legal research was revolutionized by Lexis, a widely used database of laws and legal opinions. But Codeworks didn't work.

Christopher Coyle joined Codeworks in the fall of 1989 as president and chief operating officer to help the firm with its marketing strategies.

Coyle soon found that the inherent difficulty of launching a publishing product was compounded by the softening economy and the collapse of the construction industry.

"Codeworks had many successful launches and an increasing customer base, but the reality of business is that in today's economy, you cannot afford a long ramp time," Coyle said.

"Information consumption patterns don't change overnight," he said. "It took Mead Corp. seven years supporting Lexis before they turned it profitable, and they spent $27 {million} or $28 million."

As the firm's main client base of architects and engineers began losing work, Coyle said, Codeworks tried to market its database to other users. Some insurance companies were interested in using its technical summaries to help value properties based on rebuilding costs. But it was difficult to change the focus of the business quickly enough, he said.

Bill Brenner, who joined Codeworks in its infancy and was senior vice president when the firm closed, has a different view. An architect by training, Brenner had worked in the building regulation field for years before joining Codeworks. He gradually came to realize, he said, that although the idea for a database of local and regional building codes was a good one, there were stumbling blocks far greater than the need for more time and more money.

The company's database was built on three groups of standard codes, called model codes. They are copyrighted and Codeworks therefore could not reprint them. So every code had to be paraphrased, and some tables, indexes and other highly technical material could not be used at all, which reduced the product's usefulness, he said.

Furthermore, the reports printed by Codeworks did not always allow for extensive cross-referencing, which could be accomplished with a computer disk or an on-line data service, Brenner said.

"It often meant you had to go back to the {model code} book itself," Brenner said. "If we had had an interactive product, or an on-line product, it would have worked better."

Dennis Brown, co-founder and chairman of the firm, had yet another view of the firm's fall. He said the main problem was understanding and educating users. The Codeworks database was not useful to architects doing projects in their own region, he said, because most firms have at least one employee who knows the model codes and the local amendments inside out. Those who didn't know the codes generally didn't have enough knowledge to make an informed request of Codeworks.

"Our biggest utility was being able to deal with codes in areas where the architect was not familiar with the codes," Brown said. The firm had code information for 1,000 jurisdictions and 84 metropolitan areas in 34 states, but even so, "we didn't have enough information to service all the areas where all the projects were," he said.

All three Codeworks executives, and many others in the industry, still believe that there is a market for this service if these problems are overcome and if the economy picks up. Brown said, though, "It would take someone with deep pockets who was willing to make the investment."

Brenner said that the firm spent most of the last year trying to find a buyer, which it almost did on several occasions, although nothing panned out. The company was spending about $300,000 a month on salaries and other operating costs while revenue was never more than $25,000 a month. All told, the firm spent about $9 million dollars in venture capital to get the product off the ground, Brenner said. The firm's principals still are trying to sell Codeworks's database and computer equipment.

"We were pioneers in collecting building code information," Brenner said. "What we did was a phenomenal effort. We did everything we said we were going to do except make money."