Seeing Around Corners

by Rita McGrath

  Underwriting and pricing—The core competitive advantage of many large traditional insurers was that they had access to massive databases that other companies did not, upon which they could base smarter pricing decisions. But with the advent of massive, connected, and to some extent widely available databases, plus artificial intelligence, the need for manual underwriting and the deep expertise associated with that process has disappeared for all but the most unusual cases. And proxies for favorable insurability—such as having an advanced degree or working in a certain profession—also have disappeared as decisions are more and more based on real data, not approximations.

  Claims—While companies such as Progressive have been inching toward real-time claims processing and settlement for years, widespread and inexpensive technology will accelerate this trend. Sensors, drones, and powerful smartphones will dramatically change what insurers call “first notice of loss.” Smart algorithms can then figure out what the insurer should pay, and the insured’s transactions progress seamlessly from that moment on.

  Clearly, if you are in the insurance business, you want to begin taking action to prepare your organization for the massive business model changes to come. And this is where the trouble starts. In the early days of an inflection, even though the signals are extremely strong that things are about to change, it is easy for an organization (sometimes with the help of rather expensive consultants) to conclude that they know enough to proceed, and then begin to pour money and resources into a “damn the torpedoes, full speed ahead” digitization effort.

  Unfortunately, this is almost always a mistake. Even with very strong signals, there is still a substantial amount of uncertainty about the nature and magnitude of changes to come. The task is indeed taking action and planning, but planning to learn, using a discovery-driven approach.

  Illusory Certainty and Digital Flops

  Big, ambitious digital initiatives bear a strong resemblance to big, ambitious innovation projects of other kinds. They are characterized by decision-makers having to make assumptions without the benefit of having facts to work with. Decisions seemingly need to be made—urgently—but the context is so uncertain that it would be nearly impossible to guarantee that they are correct. Instead, in this highly uncertain context, decisions are pretty much guesses. In an organizational system that is based on decision-making in highly predictable circumstances, this can lead to a vicious downward spiral.

  Decision-makers establish projections based on assumptions. As information starts to come back about the consequences of those decisions, assumptions are sometimes shown to be incorrect. And yet, having invested personal capital and often organizational capital in promoting a given point of view, decision-makers escalate their commitment, becoming even more committed to the same ideas. This train wreck proceeds until it becomes obvious that the project is never going to work and someone finally has the courage to pull the plug on it.

  Unfortunately, for those of us seeking to learn from these failed experiments, in most cases the topic instantly becomes undiscussable, the players who were involved in it disappear, and everybody remaining pretends that it never happened. This is, of course, rather frustrating for business book authors who would like to be able to derive some lessons from such failures. There are a few, however, that are sufficiently public that documentation about them survives. Let’s consider one, the BBC Digital Media Initiative, which ended up costing the organization some £98 million with nothing to show for that expenditure.

  The BBC Digital Media Initiative

  The clear specter of on-demand news and entertainment in the late 1990s was discussed earlier in the context of Netflix. Indeed, as early as 2000, it was obvious even to Harvard Business School case writers that technologies such as video on demand were likely to change the way media was consumed.

  At the BBC, the desire to do something about this looming inflection point crystallized in the form of a project called the Digital Media Initiative, or DMI, in 2008. The initial impetus for the project came from a desire to create “completely tapeless” production workflows. The BBC’s director of technology, Ashley Highfield, was the public face of the initiative, which was approved by the organization’s decision-making body, the BBC Trust, and funded to the tune of £81 million. Siemens, the BBC’s traditional technology provider, was chosen as the contractor to work on the project, with consulting support from Deloitte. The original proposal projected a total benefit from the program of £99.6 million. As a technology partner described it at the time: “The DMI is a pan-BBC project designed to prepare the broadcaster for an on-demand, multiplatform digital environment and provide a reusable foundation for the cost-effective delivery of new and emerging services.”

  Failure to Realize the Organizational Significance of a Digital Inflection

  One of the first untested assumptions that is very clear in retrospect was that this program was approached as though it were a relatively straightforward operational and efficiency-oriented project. In reality, however, such a dramatic shift in process workflows actually required a true business model overhaul, which, to be successful, would have to reach deep into the core workflows of the organization. It would, therefore, entail a significant change in the management component in addition to whatever changes the technology would demand, with substantial political battling likely.

  The contract with Siemens was awarded on a fixed price for fixed deliverables basis. The assumption of contracts priced this way is that it is completely clear what work needs to be done and what a successful outcome should be. From a user perspective, this framework had the effect of causing the BBC not to question—too much—what Siemens was working on, since they perhaps felt that interfering would negate the intention of the fixed price. From the Siemens perspective, it set up the traditional argument for why big IT projects fail: the IT vendor doesn’t deliver what the user wants, while the vendor accuses the user of changing its mind and not specifying clearly what its needs are. It is symptomatic of assuming certainty where there is none.

  The sheer number of subprojects included in the initial BBC brief was breathtaking. There was to be a new “media ingest system,” which would change the way content entered the BBC ecosystem. This would be complemented by a new media asset management system, which would be the way that future audio, video, stills, and other content would be managed. Storyboarding would be done online, rather than through preexisting manual processes. All these other systems would be accessible through a metadata sharing and storage system. The project team from the BBC awarded the contract for the work without requiring competitive bidding, and the relationship with Siemens (the primary vendor) and its other contractors was described by one observer as “distant.”

  Assuming the Whole System Must Be Built to Realize Any Value

  This brings us to a second set of untested assumptions, namely that doing these digital projects all at once was necessary and digestible by the BBC. This was a particularly dangerous idea. In an organization with low levels of digital maturity, starting small and building capability over time makes much more sense. Launching aggressively into an approach with all the funding given up front is a recipe for disaster. Even if a project works out, the organization is unlikely to have developed the new workflows and practices that will allow it to benefit. It also illustrates the mistaken idea that an organization can apply a build/development model—usually called the “waterfall method”—from a previous era to an entirely new market context. Hiring outside firms that didn’t appreciate the difference between the old and new models didn’t help the BBC project much either.

  A third set of assumptions has to do with the evident perspective of senior leadership that the BBC could safely delegate the design and implementation of such a complex system to contractors, with relatively little oversight. In a discovery-driven approach, the project would have been undertaken in such a way that it would provide proof of value all along the way. This would have involved creating quick chec
kpoints at which specific assumptions about the way the project would ultimately work would be tested, and implementing course corrections if they were thought to make sense. Instead, the DMI project didn’t appear to have an effective governance process incorporating senior business leaders as well as technical staff, or to require regular project reviews that would have fed into this process.

  Those with the Information to Sound the Alarm Had No Voice

  In violation of the principle of people with direct experience being able to sound the alarm and be heard by senior leadership, those closest to the DMI had few mechanisms to be heard. As one observer noted, “There was a culture which apparently did not allow staff involved to be given a voice, so, unable to feed their concerns about projects into review processes, they were instead reduced to privately voicing them.”

  In September of 2009, the BBC and Siemens parted ways on the project, with the BBC electing to bring it back under in-house management.

  The lessons learned from this project were bitter indeed. As a project review that ended up going all the way to the British Parliament makes clear, the assumption of knowledge where none existed could have been foreseen right from the beginning. Reviewers concluded that “the Programme had proved much more challenging than Siemens had first believed and that Siemens had lacked in-depth knowledge of the BBC’s operations. The BBC itself had only limited knowledge of Siemens’s design and development work.”

  The Right Approach—but Too Little, Too Late

  After bringing the program back in-house, the BBC staff elected to pursue a different approach, using “agile” methodologies to create some near-term positive results via collaborations between users and technologists, just as a discovery-driven approach would suggest. This turned out to be too little, too late. Dominic Coles, the director of operations at the BBC, decided to pause the project in October of 2012 and to terminate it altogether the following year. As he observed, with respect to the progress of the project,

  The pace of technological and digital change has been rapid; business and production requirements changed within the BBC; and the industry has developed standardised off-the-shelf digital production tools that did not exist five years ago. Developing such an ambitious and technically complex solution that was able to cope with the myriad demands BBC programmes would place upon it due to the variety and complexity of our content, proved far more challenging than expected, which led to delays . . . The decision has now been taken to stop the project at a total cost to the BBC of £98.4m. The cost is so great because much of the software and hardware which has been developed would only have a value if the project was completed and we cannot continue to sanction any additional spending on this initiative.

  With a discovery-driven mindset, value is created all along the development cycle, not just realized at the end. As with other kinds of innovation projects, the assumption that the entire system must be built to realize any value at all is extremely dangerous.

  Let’s contrast the BBC case with an equally big, ambitious undertaking with a strong digital foundation—the effort to bring digital insight into the treatment of cancer. As I hope to demonstrate, the use of a discovery-driven approach can yield far more possibilities for success in a highly uncertain context.

  Discovering a Working Business Model in Medical Data

  Former vice president Joe Biden’s son Beau died in 2015 from brain cancer. Biden tried to understand the medical system designed to treat his son. He was horrified. “It’s frightening, it’s complicated, it’s serious, and it’s neglected,” he said in an address to the American Association for the Advancement of Science in 2018.

  Different parts of the system were not connecting, data that could prove transformative was locked away in individual file drawers and unconnected systems, and information about what helped some patients wasn’t available to doctors who were treating others. Despite astonishing gains against the disease, the system in place to treat it had grown up haphazardly and was deeply flawed.

  In 2016, President Barack Obama tasked Biden with leading the National Cancer Moonshot, with the ambitious goal of ending cancer as we know it. This goal was deliberately intended to spark an inflection point. In particular, the Cancer Moonshot was to be the linchpin of activities that stretched across the federal government and the private sector, connecting actors throughout the system who might in earlier years never have even spoken with one another.

  At around this time, the US Congress passed the 21st Century Cures Act, which directed the FDA to permit data other than clinical trial data to be used in support of drug approvals (earning it criticism from pharmaceutical industry skeptics).

  All of that brings us to where we are in the journey of this book. The announcement of the Cancer Moonshot, new FDA regulations, and the expressed desire to create an inflection point—the messages of impending change are clear: The weak signals have become stronger. The entire arena is up for grabs. And a new way of collaborating across the whole system is clearly called for. So far, so good.

  An unlikely player to emerge in light of this massive shift in how cancer was viewed was Flatiron Health, a data-sciences software company that undertook a discovery-driven journey to benefit from the inflection point. The two founders of Flatiron were entrepreneurs at an incredibly young age, embodying the principles of entrepreneurial thinking—looking for opportunities and capitalizing on them at the right time—which entails being able to see around corners. But first, a bit of background.

  From Snake Breeding to Advertising Software

  Nat Turner and his longtime business partner, Zach Weinberg, were not your ordinary twentysomethings when they cofounded Flatiron Health in 2012. Turner was twenty-four when he sold his first company, Invite Media, to Google for a reported $81 million in 2010. He was thirty-two when he sold his second, Flatiron Health, to Roche, for nearly $2 billion in 2018.

  Although both Turner and Weinberg had been involved in the startup world prior to meeting at the Wharton School at the University of Pennsylvania, the entrepreneurship program there gave them a head start on the business that would eventually pay off in a major way. The discovery-driven growth methodology is part of the core entrepreneurship curriculum at Wharton (with credit to my coauthor and longtime entrepreneurship center director, Ian MacMillan).

  In truth, neither Turner nor Weinberg had a particular interest in healthcare. In fact, Turner’s previous businesses had begun as hobbies and morphed into cash-generating organizations in sectors as diverse as food delivery, web design, and, yes, reptile breeding. It was while Turner was running the snake-breeding business that he discovered a proclivity for the Internet. As he describes it, he made a website for himself, then other breeders liked it and asked to have something similar.

  The evolution of Turner and Weinberg’s company Invite Media is illustrative of the ideas of arena-based thinking and discovery-driven learning. The concept of doing something with advertising emerged as the two were working as interns under the sponsorship of one of the Wharton entrepreneurship adjuncts, at a company called Video Egg (now known as Say Media). While there, Turner later told the Financial Post, “we had noticed how screwed up the online advertising industry was in terms of technology complexity and adoption . . . Starting Invite Media at first was both an attempt to address that observation but also an attempt to build a big business.”

  Notice that they first defined a broad arena (online advertising) and then a problem people were having in that environment getting essential jobs done. This formed the genesis of a business that they anticipated could be very big indeed.

  With discovery-driven planning, the starting point is to define what success could look like—not to predict that it will occur, but to outline what the upside could be. Just as I proposed in Chapter 3, Turner and Weinberg began by mapping out a potential arena for the firm, charting what they called “dollar flow” (which is very similar to my suggestion to start with the pot of resources for which you are cont
esting when thinking about an arena). They then examined how badly the solutions available to different industry players did the job of matching an advertiser to the desired audience. Turner describes how the business started:

  We started as a video ad network in a way. We met the founders of YouTube before they launched and they had this idea to build this massive video destination site. We saw all this video inventory out there—all these people watching videos—and we thought, “There’s no advertising against it.” A company I’d worked at called VideoEgg had 500 million video views a day or something like that—or a month—that weren’t being monetized. It was half a million dollars in server costs just to serve these videos. There wasn’t a way to get advertisers on.

  The name Invite actually comes from the fact that we wanted to build a unit, an ad unit that popped up—kind of like when you’re watching basketball on TNT and you see the new TV show trailer pop up in the bottom left corner of the screen. That’s what we called an “invite.” It was inviting you to watch something, inviting you to do something.

  He described the initial idea for their business as “frankly awful.” They redirected the business to doing something with Facebook ads, which didn’t work out either. Eventually, they realized that there was a gap in the market for advertisers to be able to identify relevant video content to advertise against, and set about creating a program that would facilitate the monetization of advertising space.

  So the next idea was to create an electronically mediated advertising exchange. That seemed more promising, but the company “morphed,” as Turner said, into an advertising agency–based model. He said it took a year and a half to go from “Let’s explore this area” to “This is going to work, let’s hire people and build something.” As he recalled, his investors were very nervous during that year and a half, with one of them referring to the founders’ willingness to adapt when new information came in as the “idea du jour.”