The 6 Startup Failure Patterns, Why 90% Die & Jibo Burned $73M - E578

Jeremy Au breaks down why most startups fail and why it’s rarely just one thing. Backed by funnel data and battle-tested case studies, he reveals six patterns that repeatedly kill ventures, no matter how visionary the founders are. From premature scaling to bad macro timing, this talk shows how failure is often structural, not personal.

00:05 Startup Funnel Reality: Out of 1,100 seed-funded U.S. startups, only 12 reached unicorn status. Failure happens at seed, Series A, Series B and beyond.

01:26 Case Study: Jibo’s $73M Fall: The world’s first social robot died from engineering overruns, leadership disruption, and Amazon’s cheaper, voice-only Echo.

03:53 Defining Failure: A startup fails when early investors don’t get their money back regardless of user love, media buzz, or product quality.

08:00 Six Killer Patterns: Startups fail from co-founder misalignment, building without validation, misreading early traction, scaling too fast, bad timing, or relying on too many risky bets—all seen in cases like Quincy Apparel, Triangulate, Baroo, Fab.com, and Iridium.

22:40 Rebound & Revenge: Failed founders often bounce back—some become professors, others launch billion-dollar revenge startups like Rippling and Anduril.

(01:00) So I know we keep talking about failed over and over again, but what I wanna do a little bit here is talk about it in terms of stages. So this is a funnel showing that at the top of the funnel, there's about 1,100 US seed funded startups. At the end and the bottom of this funnel, 1% of them 12 companies reach unicorn status.

(01:19) So this database includes stripe, dockers, and obviously there's a range of valuations that we see here, right? Of these exit outcomes. So what's important for you to understand is that out of the thousand 100 companies that had received, about half of them failed to achieve their series A funding.

(01:39) So half of them died, right? So only 534 made it. And then at the next stage, 5 34 became children 35. So about a hundred of them. So about 20% are failed. Then about 2 35 again about. Half failed to get the next round. So all, so what I'm trying to say here is that failures can happen at every stage of the company.

(01:57) It can be early in the company, it can be middle (02:00) of the cycle, it can be late. Obviously there's a whole stack of it, but we wanna talk about why do these startup failures happen and how do we define failure as well? Because nobody wants to be a failure. And so we talked about how previously we looked at venture capital, the different asset classes we talked about, how.

(02:17) The majority investments don't work out, but only a few of them created a huge return. And so one thing we wanna think about a little bit is what is a failure, right? Are you a pioneer? So Jibo was the world's first social robot. This very cute. Little thing here was supposed to be on your desk in 2013, and then it's supposed to look at you and supposed to dance and he could make little sounds and talk to you and so forth.

(02:41) And this was done all the way back in 2013. They died after raising $73 million and they died. So they learned during engineering, there were a couple of, like mistakes you can call it, or things they had to learn along the way, but one of the many things that they learned was that engineering would be more expensive than they taught.

(02:57) For example, they thought that it could use a normal camera, (03:00) but turns out that inside the home, lighting is more dim than in the outside. And so the cameras would have be more expensive than they would have to. Another thing that, what they had to do was even if there's a camera, the camera doesn't recognize what is, the face doesn't recognize things.

(03:14) So they create middleware to be able to translate that message from a camera signal to saying this is a human face talking to me. So you're the engineer no matter where, from scratch. And of course they had bad luck on the team side. The CEO was diagnosed with leukemia. And as a result CTOs had a step up for a year as a CEO of the company.

(03:34) And of course, Amazon Echo launched out and as we know for to remember, but a lot of you will not have used Amazon Echo, but many of you now have obviously Siri. Many of you may have Google smart, speakers and it's got to the point that Amazon Echo was the first out of all of those devices.

(03:54) And the key insight that they had was, let's simplify everything. We don't need a camera. We don't (04:00) need visuals. We know a screen. We only want to have your voice, and a voice talking back to that person. So the Amazon launched that Echo speaker in 2014. And it started doing very well.

(04:11) And so they died as a company. But they were not wrong. So last year was the return of AI powered social robots. We started seeing digital companions come back 'cause of chat GPT. So 2022 chat GPT came out last year.

(04:23) Digital campaigns started coming out. Last week we saw Elon Musk launch Optimus, which was humanized robots who can pour you a drink as a bartender who can play tic-tac toe with you and hang out with you. So you know what is old becomes new. What is new becomes old. And so when we think about defining failure I think what we want to do is we wanna define startup failure.

(04:44) As a startup fail if early investors did not or never will get back more money than they put in. Okay? So this is a financial metric. And the reason why we think a financial metric is because VCs. I'm making financial decisions about where to put the (05:00) limited partners capital to work. And so they need a return, a financial reward to happen.

(05:04) So that's what we mean when we talk about startup failure. Now, I. First of all, I wanna say just because it's a startup failure doesn't mean that the founder is a failure, right? It's not a personal failure, right? A personal failure is, cheating on your spouse or being a terrible being or going to jail.

(05:19) That is a personal failure of the founder. But, so what's very important for us is we need to separate the definition. A startup failure is a bet that didn't pay off, okay? And. The fact of the matter is that VCs invest in founders knowing that 90% of the investments will be a failure. So that's the key thing that we keep reminding of is and so a lot of people say, oh, I wanna be a startup founder.

(05:44) But then you must realize that 90% of people who wanna be founders, maybe more, 95% will fail. There's a high risk job in that sense, but they'll fail in terms of the venture, but they're not personal failures. They can take those skill sets. There could have been pioneers in inventing something (06:00) new, creating a new network, creating a new technology as well.

(06:03) And the reason why it's difficult to understand startup failures objectively is because of three major reasons. The first is that we humans, we see something go wrong. We like to say one reason, and if something goes well, we wanna say one reason. So you didn't do well in university because you're lazy.

(06:20) We don't have all these multiple reasons, blah, blah, blah, these other things. People wanna simplify that, but the truth is, there are multiple reasons for happening. So obesity is a thing and there are eight known factors that can cause weight gain, right? Ranging from high stress to low sleep, to hormonal disorder, to genetics.

(06:39) And all those things happen and a lot of people be like, oh, there's only one reason. No, it's all of those reasons cause. Weight gain, similar to how there are multiple reasons why you can have heart attack, right? So there can be multiple reasons why a startup can fail, and some of them may have high weightage and some of them are low weightage.

(06:56) And The second thing is that when we have something called fundamental attribution (07:00) error, but what that means is that when we observe other people, we tend to say it is their fault and the environment's fault. When something that I do goes wrong, I tend to blame the environment and I don't blame myself.

(07:11) And our of saying this is that sometimes people say judge other people by their actions and judge me by my intentions. You know what I mean? So I understand my own intentions. I want you to judge me by my intentions, but then I wanna judge other people by the actions. you can call it hypocrisy, you can call it a differing yardstick.

(07:27) But in general, what happens is that most people, will have different reasons. The last thing that's also very difficult is that people will be finger pointing at each other. So when a startup fails. Obviously the founders will blame the environment. They'll blame their VCs. The VCs will blame the founders.

(07:42) The press will normally blame the founders. So it's just an easier story to write. And for example, if you read about SpaceX, the recent success, everybody was like, wow, Elon master this, Elon master that. But of course people ask in a very small voice like, what about engineering?

(07:58) What about his investors? All of (08:00) these people also play a huge part in it. But of course from a Twitter perspective, when you only have a certain number of words and time and tension, it's easier just to put Elon Musk and SpaceX and success together, right? In one line. But of course, it's very hard to that, and it often takes reading a biography or some sort of longer stack.

(08:19) Of knowledge in order to figure out what the truth is. With the about six clusters of startup failure this was written by my Harvard MBA professor called Tom Eman. I've tailored it a little bit as well, so I added my spin on it. So the six categories are good idea bad fellows. second is false starts.

(08:38) Third is false positives. The fourth is speed trap. The fifth is bad macro luck. And the sixth is cascading miracles, and we'll go through each one of them. So a good idea, bad partners is basically saying that it was a good idea or a decent enough idea, and unfortunately the team that came up together was just a bad team.

(08:59) It was just not a (09:00) good fit. Maybe across the co-founders, the employees, the partners, investors. Common problems that you see. One good example was that Quincy Apparel was founded by two Harvard MBA people, and basically they were inspired by Bon Nobles and other direct to consumer fashion startups.

(09:16) So obviously today, we now know Love Bonito is a success is a startup that started over 10 years ago that was Ran the Runway, which also was a direct consumer of fashion rentals in the us. Bon Nobles was successful. D two C startup brand for men's wear, primarily shorts. And so they wanted to create a direct consumer work wear startup brand for working women, which they were, right.

(09:38) They Harvard MBAs, this attire. So they felt like there's opportunity there, but they had multiple problems. I think once that problem was, they never agreed on who was CEO. They never had a final decision maker, which wrote things down. They hired employees. They felt that they could support the fact that they didn't have.

(09:55) Experience in the fashion world. So they hired people to help them, but they were (10:00) not able to support them sufficiently. And they brought on investors. they had previously invested in direct to consumer fashion, but it turns out the investors who had did that were too hands off.

(10:09) They had never really been in the guts. They had invested, but they were not, really able to support or give value and support. To the founders the way they needed it. And so they made mistakes, for example, Quincy Apparel. They obviously designed it the way they wanted it, et cetera.

(10:21) They made a production run and then they found out that the sleeves were too tight, right? So it was a basic error from the concept of fashion, et cetera. But for a company only had a million dollars of capital that's the kind of mistakes that kills the whole company. 'cause there's not enough time to recover from all of these mistakes.

(10:37) So this will be an example of a company that failed, the second is the concept of starts. So a lot of folks will get very excited about the idea and they'll start building the startup without having really done a lot of customer research or understanding of the consumer. And fundamentally, they never reach product market fit.

(10:56) So in a lean university, they misidentify the problem or they misidentify the (11:00) solution and get it wrong. So triangulate was very early, before Tinder. At the start of the internet age, and he's an engineer and he basically made the decision. He said, you know how I wanna date, I want an algorithm to decide who I should date.

(11:13) So I'm gonna upload my data from Facebook and my public information to the algorithm, and the algorithm is going to help me recommend matches for who this person should be. Imagine a screen, the UX is you should date this person because of this algorithm based on your social media posts and all, blah, blah, blah.

(11:33) Turns out, as an engineer, that was exactly how he wanted to date, but turns out the world did not wanna date the way he did. And today it is history. Most dating apps a day like. Tinder and hinge primarily start out with primarily a visual of the person the age, the distance and that's about it.

(11:52) So to simplify that, now obviously there's some spins to that, obviously some social mechanics to make it better. Some of them allow you to filter by height, (12:00) so you allow you to add your profession or your occupation or small headline to it, to have some sort of call to action headline. But it turns out that most people did not want to hire.

(12:10) Sorry, people did not want the date base on data. The third category they have is false positives. So you build the right product for the early set of customers in the early days, but then based on the wrong set of assumptions, et cetera, you end up expanding it to the wrong set of customers over time.

(12:27) And you basically accelerate and then you die because you. Grew to the wrong second step of customers. So Baru is a company Lindsay Hyde. She raised $3.6 million for PET services. Her concept is quite simple. She would partner with buildings and then she'll provide dog walking and other pet services for all of the residents of that building.

(12:47) So basically a concierge for that service. So she was successful in Boston and so she was excited to expand this to New York and unfortunately eventually failed. In retrospect, one of the reasons why it failed was that she had (13:00) expanded to Boston. And the time that she was expanding, first of all, it was during winter time, so nobody wants to walk their dog when it's winter time.

(13:08) It's snowing, it's not pleasant time. So they had more business. Two, there was a building that she had partnered with called Inkblot. And basically a lot of its tenants actually were people who are in town to do a movie shoot. So these residents were here for short term. They didn't really understand the neighborhood, and they were working like crazy on an entertainment set, and so they were very happy to pay her money to walk their dogs.

(13:35) Turns out that signal didn't translate well to other mainstream residents in other neighborhoods and other cities. This company failed. Of course, these companies that failed tend to be a little bit more under the radar. They tend to be smaller. The ones that, these are the ones that as they come along and grow up, they become more visible as failures.

(13:54) There's a company failure called Speed Track. Basically what happens is that a company tends to have very aggressive expansion. So their (14:00) product market fit, they grow a lot, they raise a lot of money, and because they saturate that early target market fit they start to expand to new markets. As a result.

(14:08) But, and the reason why I expand the new markets is because . The team and everybody does an assessment and say, Hey, there are strong network effects here. The more people use our product, the better it is. Therefore it's a winner. Takes all market, right? Where the biggest player will become the biggest winner over the long term.

(14:24) Therefore, we need to raise a lot of money. Therefore, we need to do a land grab. We need to set up flags in new countries. And if we act crazy and go very aggressive, our competitors will get scared and they will surrender to us. So that's like some of the thinking that will happen in the business side.

(14:38) Of course, unfortunately what happens is that, first of all, if you grow very rapidly it can attract rivals because they basically say, Hey, this is a really good idea. Let me copycat you. So I was talking recent podcast with Audra, but basically what she had was that. They saw the success of Groupon in America.

(14:53) So they launched their own version of Groupon in Malaysia and Singapore, which succeeded, and eventually Groupon expanded, and (15:00) they acquired basically Groupon, Southeast Asia from ra. So that was their first exit. But of course and then, next stage of course, is that, as these companies scale, they tend to be very disorganized.

(15:10) They have management complexity, they have, internal conflicts. And so there's a certain level of growth that are going very fast. And then suddenly investors get very scared about investing. And then the CEO kind of slams the bricks because they can't raise the next round. So imagine like the CEO is like raising one mil, five mil, 10 mil, 15 mils.

(15:26) They keep growing, just keep spending money, and suddenly they think they're gonna raise a hundred mil, but suddenly the last minute the investor drops out and then. They do a lot of layoffs. So fab.com did something called flash sales. They and Lazada, that in Shopee today. But if the concept was very quick, simple, there's only a hundred units of this very steeply discounted mechanic tool or fashion or 

(15:48) And so you have to be on there. And they raised $336 million to about billion dollar valuation, and then it all went to zero. Yeah, so it happens all the time. Sometimes there's bad macro luck (16:00) as well that happens. So sometimes you can figure out what the right product market fit while it's growing the customer base.

(16:06) But then bad luck happens because the economy sucks and the country sucks. The environment sucks. For example biotech in the 1990s went through a big, drought in the 1990s. Obviously a lot of people have heard of the.com bust, right? That happened in the two thousands. Clean tech in two thousands kind of Perkins made a big bet on clean tech.

(16:27) And they were right, but just 20 years too early. So 20 years ago, they basically said, the world is doomed for climate change. We need to invest in climate tech. It turns out they were right over the course of the next 20, 30, 40, 50 years. But again, VC investments can only be held for 10 years. And so investing in 2000, the year 2000 was too early 'cause they needed an exit by 2010.

(16:47) But we don't know. So today's very hot. Cleantech is very hot today in 2024. We don't know, maybe, in 50 years time we'll consider 2020 to be a tough time. We are crypto winter in 2022 after the FTX. Crunch. (17:00) So a lot of crypto startups died because they were expecting a fundraise and then they couldn't fundraise anymore.

(17:04) the macro kills those fragile startups that were short cash runway, or they were planning to raise in a certain way. Bo was a furniture startup. Basically in 2013, they raised $19 million. Basically they do what you now see on ikea.com online shipping what in America called wayfair.com.

(17:21) This is another billion dollar company, but basically they just ship you furniture online, right? this company failed. First of all, it was already difficult scaling up logistics, but then there was a bearish turn on the e-commerce bust in 2016 as well. So there. A sudden bout of pessimism.

(17:36) Another example would be biotech. So somebody was asking me about whether biotech VCs have different yardsticks for biotech. one thing that we see for biotech is that it requires a lot of sophistication, a lot of government support. And the VCs that are doing that are often thinking about the regulatory actions that will happen.

(17:53) So in the EU for example they passed legislation that said, if you are solving for a (18:00) disease. There's a genetic disease, a very rare disease, we will guarantee a certain amount of demand for drug. And so what that means is that if you are a startup going after and solving for that drug, then we hit the requirement that we need to do.

(18:14) And so what we have to do is that because there was a guaranteed give government demand for the biotech sector, for these drugs, EU startups started coming to existence. They started driving more money, feces started putting more money into that, getting sophisticated bodies, these drugs, and as a result, many of these genetic diseases and organic.

(18:33) These orphan diseases were solved for the first time. So regulatory policy can help. So biotech, but, so there'll be example of an updraft where government action lifts these startups and VCs up, but there can also be a push downwards as well. So lastly, of course, is cascading miracles. So these are probably your most famous startups your startups that have a very incredible vision, but they tend to die.

(18:58) even after raising a (19:00) ton of money, right? So what's common about them is that they tend to have to do several things. First of all, they have to persuade a new customer base to cross the chasm and do something they've never done before. Two is they probably have to deploy some kind of new technology that's come to existence.

(19:14) Then they probably have to work with the government to get it passed or legislated. And then lastly is they have to raise a lot of money to make it happen. So those are the four things that need to happen for it to happen. So we have some legendary failures that have happened before. So Iridium back in the two thousands was like, what if everyone on the internet in anywhere around the world could access cellular service and data?

(19:35) They died. Okay. Do the satellites, they launched them. They could not convince people. To adopt because at that time, in order to use a Iridium satellite, a phone would've to be this big, a giant, and it wasn't very useful versus normal telcos and stuff like that. So they died and then it was acquired by the US government.

(19:52) So those satellites, you can still see them at night, they're going by, and the US government uses them for emergency and defense (20:00) purposes. That's one. But of course we now have with starlink, which has succeeded with the exact same playbook. But now this is doing it 20 to 30 years down the road. So the technology has gotten better.

(20:12) The customers are more willing to do that. They have these people called digital nomads who wanna be in Bali, but they want to have high speed internet, so they're willing to buy starlink, for example. So there's a new customer base that is willing to buy as well. We have segue the idea was. We need to get people away from gas cars.

(20:28) We're give everybody a battery vehicle. It just happens to be on two wheels. A segway. Nobody bought it ' cause nobody wants to look lame. It failed, but in parallel another company not very far away, which less than a hundred miles away from that Tesla was invented. And the same pitch was we shall also use electric vehicles for mobility, but we're also gonna put it in a form factor of a car.

(20:49) Same technology base, same type of engineers. A lot of set engineers eventually went to Tesla, et cetera. Two different playbooks, right? And lastly, of course is web van. For those, it was back in (21:00) the.com boom, 19 97, 19 98. Basically the idea was what if you could buy groceries from home, right?

(21:07) And the company died because at that time. Every car driver didn't have GPS. So all the routes done manually. Stock ticking was done manually. People were making orders from a dial up modem. So overall it was like it wasn't much cheaper. It wasn't much faster, so it was the wrong time.

(21:24) But of course, Amazon made it happen in Singapore. We now red. Red Mar in Singapore is also considered not a success by this definition. So Red Mar did not successfully return a return of capital to his early investors. It was acquired by Lazada. And we can share that. But the founder has his own set experiences.

(21:41) It was, but today I use Red Mat for my shopping, right? And the last example we have here is Better Place. In 2007, this guy raised $850 million. His idea was we should create a battery swap network. Where electric vehicles can happen, and we're gonna partner with manufacturers to sell electric vehicles across (22:00) Israel.

(22:00) He only sold 800 cars and then a company died. So one way to think about it was that every car he sold was roughly equivalent to $1 million. Okay? So if you bought a car from him for $50,000, a hundred thousand dollars, you basically had a million dollars of VC subsidy on that car, which is crazy, right?

(22:16) But if you think about it, Gogo role, which is a Taiwanese battery swap network, has succeeded is a billion dollar company in Taiwan. And of course Tesla has succeeded in California, obviously succeeded as well because they were able to benefit from Chinese government subsidies into the battery manufacturing industry, Chinese economic miracle supply chain so forth.

(22:35) There's lots of different reasons, but what I'm trying to say here is that the difference between. A legendary miracle and a legendary flop is actually quite thin and often a function of timing. And these guys who are visionaries. The trick is you gotta be visionary not too early because you die trying to convince everybody to do it, but it can't be visionary too late.

(22:56) 'cause by then everybody knows it and so there's too much competition. So there's that (23:00) very thin line of, and these people will. Normally be considered crazy or unemployable or a really bad employee for many years. These founders as result in terms of failure even though the companies that failed may still move on with their life.

(23:14) For example, the Baruch founder failed, but she now is a professor at Harvard, MBA, and she teaches founders about, and operators about entrepreneurial failure, social teaching, and passing on those lessons. Great person. And gimme some advice along the way. And then obviously founders may choose to build new companies as well.

(23:32) They may have a rebound startup. In other words, they're very attached. So the idea of being a founder. So once the company closes and moves on, they move into another startup straight away without thinking too much about what problem they wanna solve. There's the revenge startup. So for example, Conrad Parker he built a unicorn company called Benefits, which is a benefits platform He was fired from it by the board after some press and all this other stuff. he was. Replaced by his COO David Sachs. David Sachs is (24:00) currently the co-host of the All In Podcast. He has also launched a very successful VC called Craft Ventures in New York City. And so there's bad blood between those two now, but he left.

(24:10) The first thing he decided to do was build another company that was targeting the same buyer, which is HR offices. And so he built rippling all in one HR platform. And today, rippling is a billion dollar company and Zenefit does not exist anymore as a company So Conrad Parker basically built a revenge startup and his revenge startup was funded by Y and Paul Graham.

(24:31) And I remember being in the same ecosystem around the same timeframe, and it was quite shocking because we had read the news and the news was like, oh, Conrad Parker has all these ethical lapses, blah, blah, blah. But then everyone knows Paul Graham is a very ethical guy. So why is Paul Graham allowing him to be part of Y Combinator, even though all the press is saying that he's a horrible person?

(24:51) Right now, fast forward, 10, 20 years, everybody says that Co Parker is a really good guy. He should never have gotten all the bad media articles and (25:00) everybody was into cancel culture and blah, blah, blah. But I'm just trying to say here an interesting dynamic for you to be thinking about. Another company was that if you look at Palmer Lucky, so he was fired from Oculus.

(25:12) Oculus, he created the first set. And a pioneer created these VR headsets. Some of you may already have them. I have several generations of them at home. And so they're great for VR Gring and obviously it was bought by Meta and he was fired from Meta because he decided to donate to Donald Trump.

(25:27) And he was canceled. There was a media outrage and so forth. So he was fired. And after that he spent some time, thinking about it. And he had decision to make, he had wanted to build two different companies. One company, the first company he wanted to build was in reforming the prison service.

(25:42) So in America, the prison service is a billion dollar business that keeps people and makes profit from people who are currently in the prisons. He wanted to build a company that will prevent people from going to prison. He would have a prison, but then people who left the prison and never came back, he would receive a success (26:00) payout basically for that.

(26:01) So they would lower the net cost to taxpayers. Over time, he decided it would be too difficult and he decided to build Andrew Rail which is a defense tech. So obviously at a time when he built it, it was very unpopular. Everybody thought that he was a horrible person for doing that. It's unethical to believe in defense, et cetera, years ago.

(26:18) And he was unapologetic about it. He said, I wanna make American soldiers like techno beers. We wanna have drones, anti drones, lasers, sensor towers, all the stuff that you see in your science fiction books. That's the stuff that he wants to do. And today, Andrew is his second billion dollar company.

(26:34) Okay. So those are the patterns of startup failure patterns.