Who knew there was so much in common between the rare success in mega-projects and innovation?

discovery driven planning Jun 07, 2023

Photo: The Walt Disney Concert Hall in Los Angeles, designed by Frank Gehry


Bent Flyvbjerg studies spectacular mega-project flameouts.  I study (often) spectacular innovation mis-steps.  Who knew the process that produces both has so much in common?  Who knew that the path to success is knowable?  And that we can learn from the best outliers how to do better.  Some notes from our Friday Fireside Chat last week.


Why do so many big things begun with good intentions go so wrong?


In his recent HBR article, “Make Megaprojects More Modular,” Bent argues that big-bet, bespoke projects that are custom-designed almost always fail – in other words, that failure is the norm and success is unusual.  There are, however, some fascinating exceptions such as the Madrid Metro and Tesla’s Giga Nevada plant.  In contrast, by setting out on a huge, unknown journey with lots of up-front funding, many projects unwittingly truncate learning and create the opportunity for things to go terribly wrong.  I see the same pattern in ambitious innovation programs – all the funding up-front, no time to learn and a commitment to making something happen that ultimately will not get supported by the marketplace.


Failed projects often have to be 100% complete before they add value.  They often have planning horizons extending far beyond human forecasting ability – around a 3-5 year time horizon. They are often built with a one-off mindset, which means that whatever you learn by making mistakes doesn’t help you improve with the next set of activities. And you know that you’ll always make mistakes doing something new for the first time.  All of this sounds very familiar.


Successes are the exception, not the norm


In mega-projects, as in entrepreneurship, great successes are the exception. Sure, we all hear about the ones that did exceptionally well.  We almost never hear about (or remember) the stories about the ones that never took off.  It’s also hard to tell if the founder was just standing in the right place at the right time, or if they really developed a methodology that works repeatedly over time.


For his new book, “How Big Things Get Done” (to be published in January of 2023!), Bent interviewed those rare individuals who seem to have overcome the massive likelihood of failure to achieve success after success.  Two he mentions extensively are Ed Catmull of Pixar fame, who has completely overcome the long-shot odds of the Hollywood business with an unprecedented string of successes.  We’ve got failure all wrong, Catmull recently observed, noting that we mix up the failures that are to be feared and avoided with those that are rich in learning and problem-solving.  Another person Bent interviewed is the architect Frank Gehry, whose amazing structures are iconic, and yet he’s managed to bring them to life on time and on budget with multiple projects.


The thing we don’t realize, with both entrepreneurship and big projects, is that failure is the norm.  That’s why Bent studies the rare and exceptional people who overcome the odds and why I study habitual (repeat) entrepreneurs – they’re doing something that doesn’t come naturally.  That is why we can learn so much from observing what they do.


Modular is magical


In our interview, Bent explains why doing things in a modular way leads to better outcomes.  As he says, “when you boil this down to the core, it’s about learning. And when you make things repetitive, you make them conducive to learning.  Every time you repeat, you have a learning cycle. And the more learning cycles you can do, the faster they will be.  That’s where speed comes in. The faster you do them, the more learning there will be, and you get what we call positive learning curves.”


In addition to capitalizing on learning, making your offerings modular means you can begin to generate benefits early in the projects’ life, rather than needing to wait till it is completely operational.  Many people would look at this idea in disbelief, arguing that you lose scale if you build things in a modular manner (in fact, I just had this discussion with a participant in our Columbia Advanced Management program yesterday!).  A fascinating example of why this is not true is the way in which Tesla went about building its $5 billion lithium-ion battery factory in Nevada.


The scope of the project to lower the price required for powering electric vehicles and home-power systems is breathtaking. If completed as planned, the factory will have the largest footprint in the world, a space large enough to house 107 football fields.  But in a twist on how such huge projects are created, Tesla defined a minimum viable production facility or “block” as the fundamental unit of construction.  A block could start to operate as soon as it was finished – as little as 1.5 years after breaking ground, the first in Nevada was producing the Tesla Powerwall and generating revenue.  There is a substantial backlog for the product even as Tesla adds capacity – in a modular way, of course.


As I have suggested in the past, this does not bode well for the traditional construction business, as more of its activities go modular.  I love the way Bent puts it:  “construction sites should become assembly sites.”


Doing things in parallel, rather than in sequential mode


This is a theme that is bubbling up in design circles quite a bit.  We’ve seen it in the “Agile” development movement, where the ‘waterfall’ method which lead to so much Information Technology heartache was heavily critiqued in the “Agile Manifesto.”  We’ve definitely seen it in the way agile teams in other areas go about doing their work in parallel, rather than passing bits and pieces of projects from one functional silo to the next.


And here we see the principle of using parallel work to speed up projects and reduce their risks put to work in mega-projects.  Bent’s example here is the construction of the Madrid Metro.  In most subway construction projects, a rate-limiting step is how long a stretch of tunnel a boring machine can create in a given period of time.  A further rate limiter is that most projects aren’t allowed to operate in the evenings or on weekends.


For the Madrid project, in contrast, President Manuel Melis Maynar used a radically different approach.  Rather than the conventional approach of hiring one or two tunnel-boring machines, Melis worked backward – calculating the amount of tunnel needed to meet his ambitious schedule and hiring enough boring crews and equipment to do that work in parallel with each other. At one point, there were six active boring machines working at the same time.  Note that this approach also allowed learning to grow, as the teams learned to improve through replication, analytics could be done of the most effective approaches and (amusingly) tunnel crews competed with each other to score the most tunnel bored per day as they got together at Madrid’s tapas bars!


With respect to the working days, Melis also thought innovatively about how to relax those limitations.  Rather than face opposition and objection from community groups and those opposed to the project, he involved them.  Explaining the approach he planned to take (which involved 24/7 operations), he also provided the rationale – if the community groups would support the program, they could have an operational subway system in three years, rather than eight.  Giving people that choice effectively countered opposition to the full-day working schedule, allowing even more parallel work to take place.


Tinkering around is cheap, fixing problems after launch is expensive: Think slow, then move fast


Bent says he took a lot of inspiration from the book Thinking, Fast and Slow by Daniel Kahnemann.  That book makes the point that humans use two approaches to problem-solving.  What he calls “System 1” is our quick, immediate reaction to stimuli from the environment.  It’s the immediate reaction of our brains, often resting on heuristics and stereotypes. It’s the creation of assumptions without necessarily understanding their basis in facts.  System 2, in contrast, is the process of fact-finding, testing assumptions and skeptically examining information.


Here’s the problem:  using system 2 thinking is hard work.  You have to ask “why?” dozens, if not hundreds of times.  You have to question where your assumptions came from. You have to be open to being proven wrong. You have to be willing to hear uncomfortable new information.  And so, all too often, we rush through the “system 2” part of solving a problem.


In both innovation and mega-projects, this takes the form of rushing to solutions before you have deeply interrogated the problem.  And here’s one of the most important gems from our conversation, for me, as Bent says:  “And the key thing you need to ask in in slow thinking is why why why why? Why are we doing this project? You need to spend a lot of time on that in order to find out what the goal is and then make sure that that goal is directing you all the way through. So take whatever time it takes. And planning is cheap. You know, this is just thinking and using computers and models, whatever, you know is the right medium in the area that you’re in.  This is something I discussed with Ed Catmull, the former CEO of Pixar.  This is the way to do it at Pixar.  Catmull says I don’t care how long the developers take to develop a Pixar movie. They’re actually allowed to muck around for years because it’s cheap.”


So take the extra time.  Do a lot of mental checking and engage System 2.  It can save a boatload of heartache later.


Let the data be a point of departure


When an organization calls in Bent for some advice, he has the advantage of having built an incredible database with information on the baseline rates of performance and activity in projects.  This allows him, with confidence, to say what the odds are that a given project will meet certain parameters.  The clients may not like it, but the data suggest when they are fooling themselves.


We do something similar at Valize with the construction of a reverse income statement and deliverables specification.  For instance, if you have to add $100,000,000 to the bottom line of a billion dollar company in 10 years, what does that imply for the size of your addressable market?  This almost forces you into System 2 thinking, and can often prevent misbegotten ideas from getting any traction to begin with.


Do also read Bent’s paper “From Nobel Prize to Project Forecasting” in which he describes the process of Reference Point Forecasting as a way of identifying potential project risks.


This was a great conversation – definitely to be continued when the new book is out!


Bent’s consulting firm is Oxford Global Projects.  Check out their web site for ideas on how you can escape the trap of failure in your next megaproject.


Meanwhile at Valize


Our first cohort of on-line learners are halfway through the customer insight course.  We’re getting great feedback on the course design and some folks have indicated it is changing certain decisions they were making a different way.  We’ll be running another cohort in the fall, just checking on calendars for the live on-line portions of the program. If you’d like to join a cohort yourself, or bring a team to a larger program, please contact Jacora Kiser – [email protected].


We’re launching a deep dive of a very cool looking new offering in the digital health space for a major global organization later this week.  What we do is work directly with teams to teach them to do discovery driven planning using their actual projects – so its part training, part advisory and part brainstorming.  Fun for us and very useful for the teams – we did a similar deep dive for another project for this company and they’ve been pleased with how it has advanced their innovation agenda.