Dashun Wang, associate professor at Kellogg School of Management, crunched big datasets of entrepreneurs, scientists, and even terrorist organizations to better understand the fine…
Dashun Wang, associate professor at Kellogg School of Management, crunched big datasets of entrepreneurs, scientists, and even terrorist organizations to better understand the fine line between failure and success. One surprising finding is that people who experience early failures often become more accomplished than counterparts who achieve early successes. Another insight is that the pace of failure is an indicator of the tipping point between stagnation and eventual success. Wang is a coauthor of the study in the journal Nature: “Quantifying the dynamics of failure across science, startups and security.”
Download this podcast
CURT NICKISCH: Welcome to the HBR IdeaCast from Harvard Business Review. I’m Curt Nickisch.
Long before Walt Disney created Mickey Mouse and Disneyland, he was fired unceremoniously from a newspaper. An editor at the Kansas City Star told the young animator that he “lacked imagination and had no good ideas.”
So then Walt Disney went out and bought an animation studio… but it’s not the famous one you’re thinking of. He ran that company Laugh-O-Gram into the ground. Only later did he create his Disney Brothers Studio and the animated films that defined the genre.
Failure stories like this are told again and again. An early stumble helped someone learn – and they went on to earth-shattering success. Failure is typically derided in the moment and celebrated only after the fact.
And it’s also usually depicted as black and white. When in fact, success and failure are often very close – someone might win by just a hair, or narrowly miss out.
Today’s guest has studied large-scale datasets to better understand the fine line between those two outcomes. By researching entrepreneurs who eventually go public with their companies, scientists who apply for money to run research labs, and even terrorist organizations, our guest has found insights into failure that anecdotes don’t capture.
Dashun Wang is an associate professor at Kellogg School of Management at Northwestern University. He’s also a coauthor of a recent study in the journal Nature: “Quantifying Dynamics of Failure Across Science, Startups, and Security.” Dashun, thanks for being here.
DASHUN WANG: Thanks for having me.
CURT NICKISCH: What can we learn from data about failure that we can’t just learn anecdotally?
DASHUN WANG: Data can be quite helpful because decades of psychology research or behavioral researchers have shown us there are many human biases when it comes to failure experiences in terms of the way we can recall the failure experiences and to what degree can we learn from it.
My hope is that if we can actually collect the large scale data sets that capture the ground source information about various kinds of failures as well as their successful counterparts, then can we actually put these datasets and data points under a scientific lens and to start to understand what are the signals underlying failures and success and what, is there any new relationships between them that we can uncover?
CURT NICKISCH: Yeah. As much as failure is celebrated nowadays and there’s also a culture and certainly stronger in some countries than others of never admitting failure or always putting a good light on something. And you know, never, never admit to not getting the perfect grade. Just don’t tell anybody about it and, and hide it. So I suppose you have to go to data just to actually view failure as it really is.
DASHUN WANG: Right. That was my initial hypothesis. I was just curious to see you know, does the data confirm with what we always think? You are right, I think different cultures have different propensity of sharing failure experiences. And in fact, if you think about the current, the traditional wisdom, the idea of not sharing as much of your failure experience may actually have been a sound strategy, but our new research is showing that the story is actually allows a lot deeper than it seems.
CURT NICKISCH: So tell us more about this study that you looked at academic research but also startups and innovation and terrorism. But let’s maybe start with academic research. What did you find?
DASHUN WANG: So I was really curious about as a researcher myself I feel like I experience failure on a daily basis. So one question I really – I’m really curious is just how much does it matter if you’ll fail?
In other words, imagine if you are starting out like a junior professor, assistant professor, starting out a tenure track job and if you fail, how, how much does it matter in your long-term career outcome? So this is a major question that we have been trying to get to quantitative large scale quantitative evidence for, and this is a work that’s in collaboration with my postdoctoral researcher, Yang Wang, and my Kellogg colleague Ben Jones.
So we specifically looked at, in this case, grant applications submitted between 1990 to 2005 to the National Institutes of Health for these two groups of early researchers. They’re just starting out as a junior professor. But after this event, one group will researchers – the narrow winners – had on average, $1.3 million for five years to pursue their research. But the other group of researchers got nothing. So that made me wonder then how much of a difference does this failure experience make for the near misses going for the future? More concretely, imagine the following question. If these two core people come back for a job interview five years later, who should you hire?
CURT NICKISCH: And you found that the most important thing was how people reacted after they didn’t win. If they kind of gave up or maybe didn’t apply as much for funding, that tended to have a negative impact on their success. But those who kept applying would often win grants and often end up being more successful as academic researchers than the people who went in the first place?
DASHUN WANG: Yes, that was the first surprise we got from that data. Because for a question like this, the answers used to be very simple. The idea that the rich get richer and winnings beget more winnings.
So even if the narrow winners and near misses were essentially the same people in the beginning, this very fact that one group of people won and the other group lost, this narrowing can just snowball itself to generate runaway inequality over time. So this is the conventional wisdom. So when we went into the data in the first place, we thought we will see a big difference between the two groups. Namely, the narrow winners will just become much better over time than near misses.
But when we look at their publications in the next 10 years and measure the citations of these papers, what we found is that near misses who are still here in the system by the next 10 years they somehow managed to publish just as many papers and most surprisingly, the papers they published garnered much higher impact done near winners.
CURT NICKISCH: So it sounds like it’s not just about failing but how you fail or maybe how you take failing?
DASHUN WANG: I agree with you. I think, you know, what we also find is that there are essentially two groups of people after failure, right? One group stay down, manage to stay in but the other group of people we did observe that there’s a higher attrition rate in near misses after in the next 10 years. So this means failures can be devastating in a career, it actually highlights the fragility of a scientific career.
These people usually have a very well-established and demonstrated track record for research yet what we observe is that one near miss in the early career is associated with more than 10 percent chance of disappearing permanently from the NIH system.
CURT NICKISCH: You have also looked at failure rates in startups in another study. What did you find there?
DASHUN WANG: The motivating question there is to realize the fact that you don’t really just fail once. In fact, in reality sometimes you’ll fail over and over; eventually you’ll reach your success. You know, Henry Ford, for example, failed twice before he managed to found Ford Motor, right? And there’s all the stories about J.K. Rowling who was rejected eight times or 12 times depending on which statistic you count on before she managed to publish Harry Potter.
Or the most dramatic experience I will say is Thomas Edison who failed more than 1,000 times before he landed a light bulb.
CURT NICKISCH: And he has a famous line, I think about how people don’t realize how close to success they are and give up.
DASHUN WANG: That’s right. That’s right. If I recall it correctly, it’s something like that is the, you know, many of life’s failures are people who don’t know how close they were to success. And in fact, in this new study, when we study how people fail over and over to eventually lead to success or not, we actually now have some new quantitative methods to figure out how close you were to success.
CURT NICKISCH: So what did you find from that or what do you take away from this study?
DASHUN WANG: So the major discovery of this study, a very surprising conclusion, namely a the discovery of tipping point between success and failure. You know, as people fail over and over these two groups of people can be actually quite similar in terms of their learning strategies or their characteristics.
But depending on which side of the tipping point they are in, they could actually experience fundamentally different outcomes. One group of people, if you’re below the tipping point and may be failing over and over, you try to get up and try again, but over and over you’re not learning enough to actually achieve success. This is what we call a stagnation region. You’re trying over and over, but you’ll fail to learn enough to initiate an intelligent pattern with the improvement.
On the other side of the tipping point, which is above this threshold, people fail over and over but they get fail faster and faster to eventually approach success. So this creates a quite a surprising prediction because what it means then is first of all, not all failures lead to success.
If you take these two groups of people, what you will find is that in the beginning, in their first failure, they essentially have the same kind of a performance and the same kind of characteristics. But as they fail over and over, they start to become a very much distinct two group of people because they’re in the two sides of the tipping point.
And that’s the finding that motivated us then to test this prediction systematically from three rather disparate datasets. The first data sets is the NIH grant database we mentioned earlier. Second example is the start-up domain where we look at the innovators that are involved in a startup venture and that didn’t work out, and then try again later involved in another startup venture. And maybe that didn’t work out either. And eventually you’re in a start-up venture that actually achiever initial public offering or high value mergers and acquisitions.
The third dataset, which is a quite unconventional domain, where we look at terrorist organizations. So in this case, we look at terrorist organizations that launch an attack but didn’t kill anyone. In some sense there are many goals for a terrorist attack, but one commonly used goal in the literature is to think about they try to claim casualties.
So in this case we’ll look at terrorist organizations that launch an attack and nothing happened; launch another attack, didn’t kill anyone either; and over and over eventually they launch an attack that allow them to claim casualties.
So they are obviously three very different domains. And that’s where I think I find the results quite intriguing myself, is that namely the eventually failure group, what you see is that every failure along the way, they don’t improve in their efficiencies, that in terms of time between two consecutive attempts that stays constant as you fail over and over.
But when you look at a success group, with each failure, their efficiency systematically improves the inter-event time between two consecutive attempts systematically decreases. So this means they start to fail faster and faster eventually to succeed.
CURT NICKISCH: Certainly in business there’s a lot of talk about failing fast or celebrating your learnings rather than failures. Those are kind of catch phrases you hear nowadays. It sounds like your research kind of supports that, that there is an advantage to increasing your rate of failure as long as you know you’re learning and improving from those - that actually is an indication that you’re maybe on the right side of the tipping point you’re talking about?
DASHUN WANG: I agree with you. I think not only I feel like our results support that fail fast mantra that’s very popular in Silicon Valley, but it also actually doubled down on its importance. For example, the way I see it is our results actually now suggest this idea of failing fast is not just prescriptive but also diagnostic.
In fact, if you are entrepreneur or someone that’s in the middle field over and over, you can actually diagnose yourself to think about if you are not failing faster over time, it may actually mean you’re stuck in the stagnation region, making the wrong improvement while appearing, worry, busy and productive if you are just making all sorts of unproductive improvements.
Whereas people who are in the successful region, people who are able to fail faster and faster, maybe that’s a signal that’s indicative for the way that they focus on the items that most needs to be updated while retaining all their past experience and learning productively from past failure experiences.
CURT NICKISCH: Obviously this raises really interesting questions about how you get from one group to the other or how you learn to do that, which I realize goes beyond the extent of your research. But perhaps there are some ideas that, I mean, do you have any ideas for how people can, you know, move to the other side of the tipping point or learn to do that better?
DASHUN WANG: That’s a great question. I, to be honest, this is helping my own practice as well and I feel like I’ve benefited from my own research because I often fail over and over and from this research I start to realize – it seems like the main takeaway of our results speaks to this idea of working smarter not harder.
If you think about failure, you have to realize failure is a very wonderful thing because it gives you at least two things. One is a past experience – you failed because that means you’ve actually done it, so you’ve got a lot of stuff you can resue.
The second thing failure gives you is some feedback because the very fact you failed is somebody has to tell you you failed. So it usually comes with some feedback that you can know which part – you can have some estimate that which part you did well and which part you didn’t.
Now the key here for the model to distinguish from a success region versus the stagnation region is to think about – to take this feedback and experience and squarely focus on the ones that you didn’t do as well and try to change those while retaining the ones that you actually did well and reuse those.
So it is a balance between reusing what you did well, versus creating and recreating and improving on those you didn’t do quite well. And that seems to be the critical part between people who manage cross the tipping point.
CURT NICKISCH: You just mentioned the word manage. What does this mean for managers in terms of giving teams or people second chances and helping them learn from failures to increase the likelihood of success down the road?
DASHUN WANG: That’s another great question. In fact, I’ve been talking very closely with several startups, entrepreneurs in Silicon Valley, including VC’s and others, to think about how these results may help them to identify maybe the future winners even for the entrepreneurs who are maybe today still failing.
So one idea is if you are a manager, maybe the signal we find that people who you mind sort of succeed and people who didn’t, actually very early on can be distinguished. That fact actually may have a wide range of implications for managers to think about how do we use this? Observe the way people fail over and over to actually identify today’s losers that will become tomorrow’s winners while they are still losers. And if we can do that, I think there are a lot of different potentials for this.
CURT NICKISCH: If a CEO from Silicon Valley came to you and asked you about failure, what would you tell them? Because you could certainly give them guidance on their own career and you can also give them insights that might help them with their organizational efforts. What are important things to understand about failure that, that you think more people should know?
DASHUN WANG: The first study we discussed, one sentence there I will have for the CEO is that it doesn’t matter whether your company is doing well or not. My bet is you know, over time, sooner or later you will encounter things that don’t go the way you would like it. And that’s the nature of failure that everyone’s immune to it.
And when that happens, I think it will be actually useful to reflect on our work. To think about maybe losing doesn’t mean you are out of the game and the losers in some cases will stay in the game, may actually become bigger future winners.
And the second one is to think about this idea of fail over and over. And I think there, I had a great inspiration for this research from my childhood idol. Back when I had this extremely short lived basketball career, where Michael Jordan had this very famous Nike commercial to say, you know, I failed over and over and over again in my life and that’s why I succeed. I think this is very insightful. This, at least to me is very, very insightful to think about this.
CURT NICKISCH: When you look at kind of the public perceptions of success and failure out there, what do you think is the biggest thing that people get wrong that you want to tell them to approach differently?
DASHUN WANG: You know, I was talking with my parents the other day about my recent research. And then I actually had this moment to realize just how trivial everything we talked about so far.
Because if I – I’m originally from China, I’m a Chinese native, and if you think about over 5,000 years was Chinese history, the most famous saying, if you can ask anyone, they will tell you the most famous saying is this idea that failure is the mother of success.
And it’s everything we learned growing up, everybody knows about this by the way. You know, if you think about everything we talked about, all these results, I guess what we really discovered is indeed failure is the mother of success. And I guess that’s just as we’re going through our careers and we fail more and more and we don’t share that experience. And well, we only observe other people who have succeeded. And we gradually start to forget the importance of failure, in fact, and the role of failure actually to improve our career in different ways.
So that’s where I thought was in some ways realized my, what we discover is actually quite trivial, it’s something we learned very early on as a child. But I guess over time we just thought between winning and losing, you pick winning every single time. But I think in some cases losing also can be quite beneficial.
CURT NICKISCH: So that old interview question, tell me about a failure and what you learned might be the most valuable one of all?
DASHUN WANG: Definitely don’t forget to ask that. I think it’s very easy to think about, oh, look at this person’s LinkedIn profile and the resume is all wonderful experiences and it’s easy to get – just focus on all those great achievements of this person. But I think you also want to understand how this person may persevere in the face setbacks because that’s something that’s unavoidable, unfortunately.
CURT NICKISCH: Yeah. Dashun thank you so much for coming on the show to talk about your research.
DASHUN WANG: Thank you so much for having me, it’s been a real pleasure.
CURT NICKISCH: That’s Dashun Wang, associate professor at Kellogg School of Management at Northwestern University. He’s a coauthor of a recent study in the journal Nature: “Quantifying Dynamics of Failure Across Science, Startups, and Security.”
This episode was produced by Mary Dooe. We get technical help from Rob Eckhardt. Adam Buchholz is our audio product manager.
Thanks for listening to the HBR IdeaCast. I’m Curt Nickisch.
Dashun Wang, associate professor at Kellogg School of Management, crunched big datasets of entrepreneurs, scientists, and even terrorist organizations to better understand the fine…