Selling, Educating & Philosophizing
An early stage company must be able to sell its product (or service) to a customer with one declarative sentence. The cost of selling increases dramatically with the length of a company's marketing message.
In the mid-1980s a British engineering society published a study that claimed that 85% of the life cycle costs of a product are committed within the first 5% of the design effort. The basic premise of Iconnex was that making those initial efforts more efficient would be of tremendous value to companies.
The Iconnex product, the Mechanical Engineering Workbench (MEW), integrated a drawing window, an equation solver (based upon parametric modeling), a spreadsheet and a word processor. The elegance of the product was astounding. Each of the functions was relatively immature and the Iconnex integration was truly an amazing achievement. Some of the challenges that Iconnex overcame included the absence of a graphical user interface in the operating system (DOS 3.0, I believe), the 640K-memory limitation, primitive (and expensive) graphics cards and color monitors.
At the time, conventional engineering practices limited the typical mechanical engineer to two or three design iterations before a design commitment was made. The Iconnex MEW enabled the engineer to go through 10 or more design iterations in the same amount of time. The compelling logic of our value proposition, and the elegance of our product, with a market potential of 414,000 practicing mechanical engineers, convinced us that we had a real winner. Thought leaders in the industry agreed.
Further confirmation was provided by the fact that another company, Aries, I believe, had received over $20 million of venture capital financing to develop a similar product, but on a mini-computer platform. The Aries cost per seat was going to be $35,000. Ours was less than $3,500.
Long story, short, the company consumed over $6 million of cumulative venture capital financing (probably equivalent to about $20 million in todayŐs dollars) over several rounds; never figured out how to sell the product in sufficient volumes and with adequate predictability; and ultimately sold its source code for low six-figures.
The reason? We were in the philosophy business. We were telling people how they should do business. We were not responding to their Ňtable pounding needÓ (I think I stole that phrase from Joel Adams to give credit where credit is due). We had to spend a lot of time, energy and money trying to convince people that we were right (and, by implication, that they were wrong).
Iconnex did not satisfy The Declarative Imperative.
The most impressive example of accommodating The Declarative Imperative was conceived and implemented by the founders of NeuralWare.
NeuralWare was founded in the 1980s to take advantage of the confluence of factors that would allow the commercialization of neural networking technology. With MooreŐs Law proving to be a real phenomena, computing power was rapidly approaching the price and performance that would make this practical. The vision of the founders was to develop a line of neural network products that could work on desktop computers.
Even today neural networks are not broadly known, so you can guess what it was like in the 1980s!
What follows may be more myth than fact, but itŐs necessary context. In the late 1940s and early 1950s, as computers began to demonstrate their potential, two forms of artificial intelligence were conceived: expert systems and neural networks. Expert systems can be thought of as enormous decision trees that attempt to capture the variables and decision-making process of experts in certain domains. Neural networks mimic the learning process of the brain, recognizing patterns from within massive amounts of data and adjusting over time as it absorbs more and more data. Neural networks are used today in many applications including technical stock market analysis, credit card fraud detection, process control optimization and electronic fuel ignition.
As computing power and programming sophistication improved, expert systems first, and neural networks second, became technically and commercially feasible. The Carnegie Group, an early Carnegie Mellon spinout, was one of four, I believe, prominent artificial intelligence companies based upon expert systems that was launched in the mid-1980s.
Returning to NeuralWare after that brief history lesson, you might think that its marketing challenge was even greater than that of Iconnex and even more philosophical in its market entry, but it wasnŐt!
Why? Because the founders accepted certain realities and proceeded accordingly.
First, they didnŐt have the resources to develop a deliverable product.
Second, they didnŐt have the resources to mount a missionary sales and marketing campaign.
Third, while hundreds of thousands of engineers and programmers could use neural networks beneficially, by and large, they had no need to do so. They either didnŐt know about neural networks; or knew about them, but didnŐt know if they were applicable to their situation; and even if they were, how the heck did you build a neural network?!?!
BUT out of those hundreds of thousands of potential users, there were hundreds of people in research labs, in IT departments and in university labs who were aware of the potential of neural networks. They did monitor the environment to keep track of advancements in the field. They did believe that it was only a matter of time before neural networks would be able to deliver very significant benefits.
For these few hundred people, knowing about neural networks was important. They had a need. To that end, the first revenues for NeuralWare were from the sale of a bibliography of all articles and publications on neural networking that the founders had compiled as part of their initial research before launching the company!
Instead of philosophizing to the people who could use neural networking technology, NeuralWare went to the very small and very specific market niche of individuals who were already believers and thirsting for information.
The second revenue stream came from tuition for seminars in Pittsburgh about neural networks that they marketed to the purchasers of the bibliography.
Most of the attendees from industry were at the seminars to assess the current or future applicability of neural networks within their companies. NeuralWare offered to do feasibility studies for a fee.
When the feasibility studies were completed, NeuralWare offered to develop a customized neural network system for those companies who wished to pay NeuralWare for that service.
Note the elegance of the companyŐs evolution. It SOLD products and services to people who had already been qualified as NEEDING them. The customers for the current deliverable became the target market for the next deliverable.
In the meantime, customers were paying NeuralWare to build the infrastructure and IP that they needed to develop the neural networking products that were their ultimate vision.
They recognized the power of The Declarative Imperative and used it to formulate a successful market entry strategy.
NeuralWare went on to be acquired by a firm out of Boston. ItŐs my understanding it has been taken private again and is still in operation in Pittsburgh. I regret that I havenŐt kept in touch with the founders (whom I believe are no longer with the company), but I am forever indebted to them for showing me a unique and innovative way to bring leading edge (bleeding edge?) technology to market.
á When you first start up your company, slice and dice your market opportunity until you can identify a group of specific customers who have an urgent current need for what only you can provide.
á Quite often going slowly at first is the best way to realize your ultimate dream.
á Heed The Declarative Imperative.
ICONNEX POSTSCRIPT: The prior anecdotes about Iconnex were slanted somewhat for illustrative purposes. Out of respect to the founders, investors and employees of Iconnex, I need to note that it was a noble effort that almost made it.
At one time, a major aircraft company had decided to standardize on the MEW across the corporation. They were about to issue an initial purchase order for 100 units (at a modest discount to the $3,500 unit price) and contract with Iconnex to develop client-server architecture. Between the credibility of this customer and the cash flow it would provide, this could have made the company.
As these agreements were working their way up the corporate ladder for approvals, elsewhere in the corporation, a decision was being made to decentralize purchasing, meaning there would not be any corporate-wide software standards. The 100-unit order by the corporation became a two-unit order by one division.
By the way, this is an example of SchwartzŐ Law. What, youŐve never heard of SchwartzŐ Law? Schwartz said that Murphy was an optimist!
Oh, yeah, I almost forgot.
Six months after Iconnex started, another company based upon parametric modeling was launched In Boston. Iconnex zigged. The other company zagged. That company was Parametric Technology and it has a current market capitalization of about $1.4 billion.
Damn that Schwartz!
Frank Demmler is Associate Teaching Professor of Entrepreneurship at the Donald H. Jones Center for Entrepreneurship at the Tepper School of Business at Carnegie Mellon University. Previously he was president & CEO of the Future Fund, general partner of the Pittsburgh Seed Fund, co-founder & investment advisor to the Western Pennsylvania Adventure Capital Fund, as well as vice president, venture development, for The Enterprise Corporation of Pittsburgh.