The Internal Startup Model
Innovation and optimization are opposite cultures of product development that should not be mixed. This article evaluates the most common approach used to keep them separate: the Internal Startup Model.
The usual way to create separation is to carve out a separate R&D team comprised of people who specialize in innovation culture that report directly to the CEO, are incentivized differently, and housed in a different building (preferably). Another way is to use consulting firms that specialize in innovation, such as IDEO.
However, separation is not enough because, by itself, it carries major side-effects that are as likely to torpedo your attempts at innovating as trying to combine it with the activities of your core teams. I believe this is a surprise to many executives and goes against some traditional thinking so I’ll try to explain why.
Can’t Explore the ‘Adjacent Possible’
One of the biggest reasons that innovations fail is that they are too ‘far away’ from what is practical now by the company. For an innovation to be successful, it needs to be within a few steps of what the company and its market currently know, in a field of possibility that evolutionary complexity theorist Stuart Kauffman calls the ‘adjacent possible’. Facebook, Snapchat, Instagram — these are all services that while distinct and disruptive are but a few functional twists away from each other (e.g., ephemerality, posts, photos). Great innovations nearly always have that distinct characteristic of being only a tiny bit different from things that have gone before them. That one difference makes an outsized contribution that turns the idea into something completely new. We lose this perspective in the retrospective hero narratives that are told after an innovation becomes successful — like the ‘Missing Link’ in human evolution, the trail is lost and the surviving line looks like it sprung out of nowhere. (See The Arrival of The Fittest by Andreas Wagner on the power ‘adjacency’ in gene networks, or Stuart Kauffman’s masterwork, The Origins of Complexity, or the wonderful synthesis Where Good Ideas Come From by Steven Johnson).
I believe this to be the hardest quality of any R&D team or outsourced project to capture, but it is the one most critical for getting the right innovation adopted by the company and its customers. It usually requires people who are intimately familiar with the day-to-day technologies, customers, and problems of the optimization culture, who are then allowed to play serendipitously, in a bottom-up fashion, along branches of product-development that shoot out from the main product pipeline.
Teams that are too separate from the core business are not able to do this. Their work is either too directed top-down by executives signing the big contracts, or the engineers do not have the type of intimate knowledge of the parent business to generate ideas that are close enough to be hits.
Misleading VC Model and Survivorship Bias
When executives think about investing in innovation, they find it irresistible to think of themselves as venture capitalists. They imagine that they are investing in a venture the way a VC does. Employees are also attracted to his model, who imagine themselves as entrepreneurs — an impression that the executive often encourages. But nothing could be farther from the truth. A VC is placing bets on top of a vast ecosystem of risk. They pick and choose ideas from a playing field where most the failures have already been weeded out, the ideas a farther along, and a lot of information is known because the entrepreneur has had to toil for months or years before getting their idea in front of the VC. Even after this process of natural selection, the probability that any one of their investments will pan out is still small. Unless you are Google or Facebook (and maybe even then), it would be suicidal for companies to try to contain that ecosystem of risk within their own operations — they’d go bankrupt in 5 minutes. And yet they talk that way when they describe how they invest in innovation.
I believe the source of this delusion is a common cognitive bias known as “survivorship bias” — calculating the odds relative to the visible companies who got funded rather than the true base of everyone who started a company.
This results in executives investing far too much in far too few ideas at far too early a stage in their development and abandoning the idea when it doesn’t immediately pan out. It’s a common scenario.
Lack of Ownership
To have the best chance of success, you want an innovation to be championed by the people who will be implementing and managing it long term. But if you plan to integrate the product into your core company, the internal startup model will create too much separation between the hands that invent and the hands that implement, resulting in the new initiative being imposed on the parent company as an orphaned, foreign idea. Conversely, the inventors won’t feel like they have skin in the game or enough upside to make sure it succeeds and will start to behave like consultants.
Insufficient Diversity of Ideas & Poor Selection Process
Even with a large, focused portfolio of really good ideas, only about 1 in 20 will be worth moving to the prototyping phase. This puts a premium on the breadth of the top-end of the ideation funnel.
But there usually isn’t enough creativity in a small team modeled on a startup to generate enough ideas of adequate breadth and depth, and there is too much attachment to personal ideas which also limits creativity. A startup, after all, tends to be a company with only one big idea. And because of the investment model implied by the startup model, the process of whittling down the ideas to select the best one to go after has to be done before the team is formed. Instead of an ecosystem of innovation risk that only surfaces ideas with real-world traction, you have people with MBAs evaluating business plans. MBAs are all creatures from optimization culture. This often results in the executive team picking something they champion because nothing has gotten traction yet.
Inflated MVP (Minimum Viable Product)
We talk about the Minimum Viable Product but we struggle with what the “minimum” is relative to. One reason is because the startup is a model for a whole company. Most of the time you should only be testing a product hypothesis. Even the concept of ‘product’ is too big for many innovations. Bundling the hypotheses into products is a secondary step. Minimum means the smallest anything that can prove a core hypothesis about the product.
Inadequate Definition of ‘Success’
Startups do everything and anything to be successful, but they don’t actually need to have a specific definition of ‘success’, other than to create something that will have product/market fit and provide a return for their investors in a time-frame they are prepared to accept. They fail when they can’t do that in time — before they lose the ability to persuade enough investors to continue to fund them, and run out of money.
This is not a good model for corporate innovation, because lacking objective success criteria and clinging for life, your internal startup will become a political football every time its funding comes into question. This is generally a lot more frequently than in a real startup. The ensuing paralysis will mean that your company will either shut off the project prematurely, or waste resources funding the idea well past the point where it could have been efficiently and objectively evaluated. It’s common for businesses that have been around a while to maintain many zombie products that aren’t quite dead enough to warrant complete retirement, cluttering up the user experience and dragging on resources, accelerating the decline and making it harder to innovate.
In the next article, we will discuss a more promising model for solving the incompatibility of optimization and innovation: The Product Discovery Model.
For a deeper dive on the organizational culture differences between innovation and optimization, please refer to my previous article in this series, The Culture Clash of Innovation.
