Following Tolstoy, “All successful startups are alike; each failed startup fails in its own way.” Some startups fail because of flawed product execution. Some fail because of poor marketing. Some fail because they can’t get the right people. Some fail because they run out of money. Some fail because they did not understand the marketplace and the competition.
This story is about an anonymous company that failed for a relatively unique reason: a yawning gap between vision and product concept.
You might think that vision already implies a product concept. Yet actually the same vision can be realized via many product concepts. This particular company’s vision was to automate enterprise knowledge management and distribution. A wonderful vision, but one which could take the form of many different product concepts. For instance, one concept would be to build a monolithic, closed, smart knowledge management system. A very different concept would be to build smart knowledge management widgets that fit into and complement existing knowledge/content management approaches.
This particular company’s failure began with being funded–over-funded, actually–on the vision alone, with no-one from the founding team, the investors, or the board able to delineate the actual product concept, or more importantly, even being aware that the product concept needed to be delineated. Ideally drivers of the product concept would have included development feasibility as well as marketability, in particular synergies with existing knowledge management stacks. Making such crucial judgments requires “full-stack managers” with an unusual ability to gaze across the technology and market landscape. Yet this company, at the core of which needed to be a deep understanding of smart knowledge technologies such as categorization, targeting, and NLP, had no founders with anything but large-company product management experience, who brought with them plenty of experience in spreadsheets and PowerPoint presentations, but neither basic software development expertise, nor advanced expertise in AI or NLP, nor insights into knowledge management. Their assumption was that people with these specialties could be found and bought for the right mix of salary and equity. But those people were brought in too late, and not given enough of a voice, to actually influence the product concept. In their absence, the product concept defaulted to being a last-decade-style, big, stand-alone content management system with some under-powered smart knowledge features.
But the enterprise sales cycle for a content management system is measured in years. For an enterprise CIO, the buying decision comes down to a trade-off between keeping their existing content management systems, with hundreds of thousands of existing documents, trained users, and deep integration into the client’s basic workflows, versus migrating to a new, untested, supposedly better one–although many of the advertised features remained mere vaporware–which lacks parity with the existing systems, and inertia inevitably ends up driving such decisions. Selling to these CIOs requires armies of mid-six-figure salespeople.
The correct product concept, of course, is one of smart widgets, or bots, or agents, which roam through existing knowledge bases and usage patterns and serve up insights, recommendations, and management information. The irony is that such a product concept could both have been implemented at a fraction of the effort of the content management system, and been much more easily marketable. The company might have had to give up on its fantasy of becoming a $10B company competing with Oracle selling $500K contracts, and getting to a big IPO that would have gotten the CEO his private jet (or, in one notorious statement, gotten the since-departed CPO his $20M trip to outer space); on the other hand, the company could conceivably have sold thousands or tens of thousands of smart knowledge bots at $25K each and still reached revenue levels of hundreds of millions of dollars, with a valuation of $1B or more, not to mention a much easier path to being acquired.
But there was no one at the company with enough insight, or influence, to sell this product concept. Even if there had been, as one approaches the end of the runway, there is a tendency towards the desperate Hail Mary pass–“we have three big deals that are real close, they’re going to close next quarter, and we’ve promised the board they will! All the clients want customizations and free migrations, so we have to spend all our energy doing those! We have three customers running trials and we can’t let them down!” One can only pivot when there is enough room left to pivot.
Only partially orthogonal to this phenomenon was a very clear attitude on the part of the founders that it was their company and they knew what they were doing and they would make all the decisions. The founders’ rock-solid certainty that they knew exactly what they were doing and had all the answers also extended to the faulty product concept. A variety of people in the company at a variety of levels did in fact speak up about the product concept and related issues, and were shut down not on the weakness of their ideas but essentially on the basis of, we’re smarter than you.
Since I live in India, let me give the Indian spin on this story. Our anonymous company had deep Indian connections in their founders, investors, and board. It was thus an obvious element of the strategy to set up a development center in India where costs were half those in the US. In fact, they set up a large Indian development center which at its peak numbered more than thirty. Yet one must ask what the mission of the Indian development center was supposed to be beyond simply churning out lines of code. In this case, the very existence of the Indian development center in fact promoted a mentality of churning out these lines of “cheap” code. The Indian development center was never charged with any overriding mission or high-level deliverables. Its availability in fact served to perpetuate the failed product concept of a big me-too content management system, since any desired feature could always be produced there “cheaply” in the next sprint or two. The problem was compounded by the fact that the manager of the Indian operation had neither any meaningful voice in management nor the expertise to drive any technology breakthroughs.
What are we to take away from this? In a company whose differentiator is technology–in this case, knowledge-related AI/big data/NLP technologies–the CEO, or those very close to him, preferably on the founding team, and in any case within the top group of decision makers–must have a broad, compelling, cross-disciplinary vision of the market and technology landscape and a crystal-clear, realistic understanding of not only what the company intends to accomplish but also how. He must empower his team at every level. He must also be ready to pivot at every turn. As for the board, it must hold his feet to the fire, including requiring that he provide a rock-solid scenario for how the company’s products fit into existing ecosystems and can be realistically and reliably monetized. If the company is prematurely over-funded, as this company was, it must operate as if it had one-fourth the money it does and plan to reach break-even or true next-roundability on that basis. The product concept, not just the vision, must be the subject of intense deliberation and analysis of alternatives.