00:00:04 Olivier Ezratty’s intro and Microsoft experience.
00:01:32 Topic: Reality vs. myth in startups.
00:03:38 Overpromising and trust in startup-corporate relations.
00:05:01 Dynamics and pitfalls in supply chain.
00:06:40 Big data, deep learning in forecasting.
00:08:00 Misconceptions in startup funding.
00:08:22 Case study: Overfunded startups, Theranos and Magic Leap.
00:09:35 FOMO-driven investor behavior.
00:12:29 Need for due diligence in investments.
00:14:18 Importance of tech and market knowledge.
00:16:00 Role and challenges of innovation in scaling.
00:16:37 Debate: Company innovation understanding in market.
00:17:56 Role of AI in companies.
00:19:41 Complexities of supply chain optimization.
00:22:02 Myths about forecasting in supply chain.
00:23:55 Assessing startups: team, skills, uncertainty handling.
00:25:02 Entrepreneur’s balance: being visionary vs. hands-on.
00:27:13 Advice for entrepreneurs; Joannes’ Lokad journey.
00:29:57 Addressing ‘uncool’ problems: waste and recycling.
00:30:52 Startup challenges: service to product transition.
00:32:31 Closing thoughts.
In an episode of Lokad TV, Olivier Ezratty, author of “Guide des Startups”, and Joannes Vermorel, founder of Lokad, discuss the reality versus myths of startups. They touch on the struggles startups face, including the pressure to overpromise, stretched resources, and long sales cycles, particularly in B2B software. The conversation explores how large companies externalize risk through partnerships with startups. It also highlights the consequences of misconceptions about funding and product delivery in entrepreneurship, the impact of investor FOMO on funding strategies, and the need for technological literacy among investors and entrepreneurs. The discussion concludes with advice for startups, underscoring the importance of problem understanding, customer focus, and disciplined product development.
In the latest Lokad TV episode, host Kieran Chandler is welcoming two technology industry veterans, Olivier Ezratty, the author of “Guide des Startups,” and Joannes Vermorel, the founder of Lokad. They are discussing “Myths versus Reality in Startups.”
Ezratty is beginning by sharing his journey, transitioning from a software engineer to marketing at Microsoft, and finally becoming a business angel in 2005-2006. Throughout this transition, he observes the world of venture capitalists, key success factors, and money flows, leading him to write “Guide des Startups.”
Vermorel is joining the conversation, acknowledging the challenges of running a startup: stretched resources, constant pressure, and a product that’s never ready enough. He observes that these struggles often push startups to overpromise.
Speaking about B2B software, Vermorel explains that the situation can be complex due to long sales cycles. He emphasizes Lokad’s commitment to staying truthful to their abilities and promises, regularly discussing trials and errors on their blog.
As the conversation progresses, Vermorel questions Ezratty about startups’ buzzwords and “magic pills” or disruption claims. He is curious how startups manage to remain truthful despite these constraints.
Agreeing with the tendency of startups to claim more than they can deliver, Ezratty justifies it by calling startups “dream companies”. He believes that real innovators are those who change the marketplace and take risks, testing various options simultaneously. He discusses how large companies interact with startups, suggesting that the true motive is externalizing risk.
Sharing his observation on startups, particularly in B2C software, Vermorel notes that they often exaggerate their capabilities, leading to client mistrust. He highlights how Lokad initially dreamt of creating sophisticated forecasts using multi-industry data, but the reality required a greater emphasis on having more data.
Ezratty is addressing entrepreneurship myths, including easy funding and timely product delivery. He identifies different funding sources for startups, like public funding, venture capital, corporate ventures, and the emerging field of cryptocurrency.
The discussion turns to companies that have taken excessive funding but failed to deliver, such as Theranos and Magic Leap. Despite these failures, they examine success stories like Facebook that justify high-risk investment strategies.
Moving on to technology understanding, Vermorel is highlighting the importance of due diligence, given the vast amounts of money invested. Ezratty is expressing concern over the misunderstanding of AI in the entrepreneurial world, urging for a deeper understanding among investors and entrepreneurs.
They discuss some common myths in the supply chain industry, primarily the overemphasis on human intuition over statistical or AI-based forecasting. Ezratty points out that many companies have significant data but unclear goals for its use, emphasizing the importance of considering market dynamics, competition, consumer behavior changes, and technological advancements.
Transitioning to startup evaluation criteria, Ezratty stresses the importance of a competent, listening team, entrepreneurs’ ability to balance long-term visions with short-term management, and the quality of the startup’s proposed solution. Vermorel advises startup founders to focus on fundamental, stable problems rather than transient issues, emphasizing the potential in tackling “uncool” problems that may be undervalued.
Wrapping up, Ezratty highlights the challenge for startups to create a product rather than a service, requiring a unique mix of understanding customer needs, business scalability, and technological feasibility. He advocates for learning these critical skills to succeed in the startup world.
Kieran Chandler: Hi there, welcome back to Lokad TV. Today I’m delighted to say that we’re joined by Olivier Ezratty, who has over 30 years experience in the technology industry including 15 years at Microsoft. He’s also the author of the “Guide of Startups” which is now entering its 22nd edition here in France. Olivier, thank you for joining us here on Lokad TV today. Perhaps a nice place to start would be with just a bit of an introduction to yourself. How did you first get interested in startups?
Olivier Ezratty: I started getting interested in startups about thirty years ago. But first and foremost, I’m a software engineer. Before doing marketing at Microsoft, I was a software engineer for about four or five years. After being a CMO, managing various relationships, and starting the startup relationship ecosystem within Microsoft, I thought it would be interesting to bring back some skills to startups in the French ecosystem. So, I became a business angel around 2005 and 2006. I had a couple of small companies which helped me understand the world of VCs and the key success factors. My goal was to learn and share, which is why I wrote this guide and have continuously updated it since then.
Kieran Chandler: That’s fascinating. As always on Lokad TV, we’re joined by Joannes Vermorel, the founder of Lokad, who might know a thing or two about startups. So, our topic today is ‘Myth versus Reality in Startups.’ Joannes, what do you mean by the myths we’re seeing in startups?
Joannes Vermorel: As a startup entrepreneur myself, I can say that we’re always facing difficult situations. Our product is never ready enough, and we never have enough time or funds. So there’s always a stretch and an urgency to push something to the market. This puts us in a situation where the incentives to stretch the truth are strong, often presenting something that’s a bit more than what we can actually deliver. In the case of B2B software, the situation may even be worse because the sales cycles are very long. So, one could stretch the truth, and by the time the deal is closed, the company might have actually had the time to deliver what was initially promised, due to the long sales cycle. We’ve tried to stick very close to what we’re doing at Lokad and have discussed extensively our trials and tribulations in our blog posts. However, I find it interesting to hear from Olivier, who I consider one of the greatest experts on the French startup ecosystem. Olivier, what is your opinion on these various areas, buzzwords, and startups trying to claim their own ‘magic pills’ or ‘disruption factors’? Do you agree that startups tend to promise more than they can possibly deliver?
Olivier Ezratty: Yes, I agree, startups often claim they can do more than they can. They need to do this because a startup is a dream company. They dream about the future, about creating things that don’t necessarily exist yet. We know there’s a high failure rate, so an entrepreneur who isn’t pushing the envelope far enough is not a real entrepreneur. The real innovators are those who are changing the marketplace, and this involves risks. You need to test various options simultaneously. Large companies often work with startups as a way of externalizing this risk. The myth is that this will help the large companies to innovate, but the reality is they continue to innovate internally.
Kieran Chandler: Too heavy, too complicated so they just ask other people to take the risk and get the burden out of it. Okay, another question. Are there other incentives or reasons why startups might not be entirely truthful? Joannes, what are your thoughts?
Joannes Vermorel: Yes, one of the points I see, especially in B2C software that deals with complex systems like supply chains, is that significant innovation requires several years of effort. Startups may exaggerate their capabilities, which can lead to a dysfunctional relationship between the startup and the company trying to adopt the technology. This mismatch can lead to failure for the wrong reasons, such as trust being destroyed before there’s time to polish and perfect the system.
I’ve seen this especially in supply chains where there’s inherent complexity, dealing with many countries and systems that have to interconnect. In terms of B2B enterprise software, it’s complications on steroids. Large companies often jump from one potential startup solution to the next every two years, but always seem to fail two years short of being able to roll out the solution.
This situation creates a kind of AI winter, where companies try and declare something a failure prematurely because it was taking more time than expected.
However, on both sides of this equation, there’s a need to dream and imagine a lot of things. For example, early on at Lokad, we had the idea of having a forecasting model that could leverage many data sets coming from various industries. The thought was that this would allow us to tune our forecasting model better.
We eventually managed to implement this, but it took eight years and ended up being different from the initial concept. The initial idea was to capture early trends from fashion to predict consumer electronics consumption. That didn’t work. What did work was leveraging data from different verticals to improve forecast accuracy, mostly by having more data for the deep learning gradient descent, which made it more stable. This allowed us to use more parameters and ultimately improve performance, even without domain-specific transfers of information.
Olivier Ezratty: The common misconception in entrepreneurship is the idea of easiness. There’s this myth that it’s easy to get funding, easy to acquire customers, easy to deliver a product on time. Most of the time, this misconception is linked to a lack of experience. When someone is fresh out of school and starts a startup, they lack experience and tend to be overly optimistic. They’re trying their best, hiring the best people they can.
Kieran Chandler: Let’s talk about funding. Where does that funding come from? Who’s actually funding these startups?
Olivier Ezratty: There are several sources. In France and some parts of Europe, there is public funding. There are also venture capital firms, Initial Coin Offerings (ICOs), and corporate venture funding. ICOs are a bit uncertain; usually, they’re backed by Bitcoin value from individuals who have invested in bitcoins and choose to invest in blockchain companies.
Corporate venture funding has seen a sharp increase in the last three years, which didn’t exist ten years ago. For instance, SoftBank from Japan has raised more than three hundred million dollars or Euros, with significant portions coming from other corporations like Samsung, Total, and others.
Kieran Chandler: Joannes, what are your thoughts on this? We’re discussing instances of companies pushing the envelope, and there are quite a few examples in the real world where companies have perhaps pushed a little too far. They’ve taken on excessive funding to develop technology that hasn’t quite worked out. I believe you have more expertise in this area than I do, so could you provide some notable examples?
Joannes Vermorel: Sure, one example that comes to mind is Theranos. This company went too far with way too much money. In the end, they took somewhere around $2 billion, which is a significant amount.
Kieran Chandler: This story is indeed remarkable, and most people know about it. It was a healthcare company, a medtech company, and the founder, Elizabeth Holmes, had a dream to produce a product but no concrete plan on how to make it. The idea was some kind of cheap blood testing thing. She managed to gather funds from investors who were not very knowledgeable about this market, including US politicians and big names like Henry Kissinger and James Mattis. However, despite her efforts, it ended up becoming a scam.
Olivier Ezratty: What’s interesting is that there was a French journalist who wrote a paper in the Wall Street Journal about this scam two or three years ago. Despite this expose, the company managed to double its funding from around $762 million to $2 billion even after the scandal started. Another less known example is Magic Leap, which gathered $2 billion for an augmented reality headset, though it’s uncertain if it’s worth that investment.
Kieran Chandler: So why are investors so eager to fund these ventures?
Olivier Ezratty: It boils down to the Fear of Missing Out. Large investors, particularly in the U.S., want to be certain they don’t miss out on investing in the next Facebook or another successful worldwide company. So when they spot a company with the potential to disrupt an entire sector like healthcare or transportation, they invest heavily in it. They want to send a message to other investors that they need not invest elsewhere. It’s a kind of a war, a signal they send to others. Sometimes, it fails, as we’ve seen in the two examples mentioned. However, sometimes it works. Facebook, for instance, raised about half a million with Russian money before their IPO, and it ended up being a success.
Kieran Chandler: Joannes, how can investors improve the clarity of what they’re investing in?
Joannes Vermorel: That question reminds me of what Olivier has been doing through his blog, which I’ve been reading for a decade. Olivier extensively surveys the landscape of buzzwords like AI, blockchain, quantum computing, genomics, etc. Though I am running Lokad as a side business, I occasionally perform technical due diligence missions funded by venture capitalist firms, who ask me to conduct a technological audit of some software companies. However, I’m still puzzled by the apparent level of amateurism regarding the amount of due diligence, which seems mismatched with the sheer volume of money being invested. It sometimes works out because the small odds of success are sufficient for it to pay off, but there’s certainly room for more rationality in the market.
Kieran Chandler: It’s like magic that you cannot figure out, what is the physics behind it, you know? What is your perception on educating the market at large and embracing these topics, rather than say, “I’m going to trust the experts?”
Olivier Ezratty: It’s tough because we see different topics popping up every year. Blockchain is quite new, for instance. We’re facing more and more complicated topics, so you need more expertise and more time to get through the understanding of all these new techniques. AI is part of a number of misconceptions. For instance, people who think that deep learning is doing everything, when it does only 25% of what you can do with AI. There is a lot of misunderstanding around because people lack knowledge.
The world of entrepreneurship, and I’m not sure about variations between countries, but at least in France, is a mix of engineers and science people, but also a lot of non-scientific people. They often don’t have any clue about all of this, leading to a lack of understanding of science for many of these topics. I know a lot of companies were created based on AI by people who don’t know anything about AI. They think they can just do something, so they say, “Ok, I’ve got an idea, we’re going to create a chatbot for whatever thing they want,” and then they hire some people but they don’t know if it’s possible to execute.
Take Elizabeth Holmes for example, she was from Stanford, had only one year of graduation in healthcare, and then she said, “we’re going to do blood testing.” She had no idea. It’s kind of crazy. People create things but they don’t have enough science background.
What I’m trying to evangelize in the market is: elevate your understanding of science. You need to do that for two reasons. One, if you’re an investor, you need to be able to do the due diligence process of these companies. And second, you’re going to understand what’s happening. You’re going to have a clue of what you can do in innovation. I think it’s going to be very useful to create worldwide companies.
If you think that you are doing just an intermediary on-site business, it’s very difficult to scale worldwide because American companies are going to have more money than you. You can’t easily create a Facebook based in France. However, you can create a global company if you have some technology that can disrupt, where there’s some magic inside the technology that nobody knows about.
Kieran Chandler: Well, what I’m really getting here is this: It’s not really in a company’s interest to improve this kind of understanding in the marketplace if they’re making money and being invested in, is it? Why is it in their interest to improve the understanding?
Olivier Ezratty: Well, it depends on where you are in your product lifecycle. If you’re creating a new product category and you need to educate the market, maybe you need to explain a bit about the inner workings of your products and your technology. If you’re a leader and don’t have a lot of competition, you can protect some of your IP with industrial secrets. So you don’t explain how it works, it’s a magic trick.
But if you have a lot of competition, if you’re not leading and have some differentiation, but you need to explain it, then you need to explain where the stuff in your product comes from. This is very interesting because, with AI, many large companies went through these stages. At first, they were publishing a few research papers, Google did that a lot in the very beginning to get traction and hire more people. Then they went very strong on their secret sauce and then, at a more mature stage, they start publishing again because there are so many competitors that they just want to literally win the mindshare battle, hearts and minds. They want people thinking about them, building around their product and using their processing units and everything.
Kieran Chandler: There’s a big misconception about AI. Most of the science behind AI is public. It’s in the public domain. You can find anything in research papers. I think probably only two or three percent is missing. But you need the skills to understand the tools and then apply it for a solution. The knowledge about AI in a start-up is about how to assemble all of this. Another misconception is that AI is a product. It’s not true. AI is a toolbox with a lot of tools.
Joannes Vermorel: It’s like Lego with all these different pieces. You say, “I create a 2D dinosaur, or I create a space shuttle.” But it’s going to be complicated. The skill with AI is how do you assemble all those bricks, like machine learning, deep learning, natural language processing. That requires a lot of knowledge, and the integration requires a lot of knowledge. People think that it’s magic and that it has a lot of value.
Then you need your data. You need to update it, you need to check the quality of the data. That requires a lot of knowledge, and then you need to know the business of your customers exactly.
In the specific case of supply chain, there is an extra twist. You need to define what you optimize. You’re extracting data from your ERP or your company systems to do some kind of optimization. But you don’t just want to optimize percentages, you want to optimize results. You have to write down the formula of what you’re optimizing. For most of our clients, it’s the first time in their history that they have to have an explicit financial optimization.
The problem is that you can do it wrong by being very short-sighted. You need to think of a formula that reflects your true strategic mix and not just short term objectives. For example, if you want to optimize the price in the store, a naive statistical analysis will tell you any store in Paris can raise the price by 20%, and your margin will skyrocket for a couple of weeks. But then people will go somewhere else because they will learn that you’re way too expensive.
Kieran Chandler: Let’s talk about the supply chain industry. What sort of myths are you seeing in the marketplace that other companies are putting out there?
Joannes Vermorel: One myth, specifically for Lokad, is that there is something extremely specific about the human mind when it comes to forecasting the future from the supply chain angle. Imagine you have a company with 100,000 SKUs. Most of those products are sold intermittently; it’s very erratic, super noisy.
Even though we are a couple of decades into having statistical methods to do this, we still have a lot of people who are non-believers in statistics. And the reality is that there were a lot of startups that did pretty bad statistics, which is even worse than a human who is just approximately correct. When you do bad statistics, you’re just exactly wrong, which is kind of even worse.
Also, we’ve had several waves of innovation that just added complications. For example, big data. A lot of companies in supply chain have a lot of data. They moved to big data systems but not really with very clear goals on what they want to do with that. So they ended up having a lot of Hadoop clusters with fuzzy purposes.
Olivier Ezratty: I agree with what you’re saying. AI is not a product, it’s part of the toolbox. Big data was kind of the same. What I see more generally, not specifically in the supply chain sector, is a kind of biased mirror effect when you use your data. It’s data from the past, but it has to be used properly.
Kieran Chandler: Most companies want to predict the future with data from the past, but there’s a danger with that. It’s like driving and looking at the rearview mirror—you might not see the tree ahead and then you hit the tree. So, let’s say you take France’s Canal+. Perhaps they do some surveys, but they have Netflix. And when Netflix was released in 2014, they said it’s going to be easy to beat them. Now, Canal+ has stopped doing VOD and Netflix has the market. So, it’s interesting to see that if you don’t have good product marketing in your company and you just believe in data, you don’t look at the competition, you don’t see how people’s behaviors are changing with new technologies and services, you miss the whole thing and the data won’t tell you. So, how would you assess that startup, Joannes, and how would you assess, Olivier, a company such as Lokad?
Olivier Ezratty: In the supply chain or generally, there’s no magic trick. I look at everything. First is the team. Who are they, where are they coming from? Are they good people? Are they listening? The listening skills of an entrepreneur are very important. In a sales call, a good person is listening more than talking. But that aside, it’s very important to understand, to listen. And one of the things for an entrepreneur is managing their dual nature. An entrepreneur is somewhat of a schizophrenic, because they have to dream big and aim to change the market, but they also have to keep their feet on the ground. They have to understand their P&L, they have to hire and manage people, reward them. These are very traditional management tasks. So, the balance between long-term, short-term vision and hands-on management is tough. If you talk to an entrepreneur, you can see that in their psychology. You see if they’re able to move back and forth from these two dimensions.
The second is the idea. There are so many bad ideas in the ecosystem. You go to a startup fair, out of 1000 companies there, I’d say 80% are bad. So you have so many bad ideas, even with good teams. Some investors say, “Okay, it’s a bad idea, but the team is good, so let’s go.” I say no. You need a good idea and a good team. So, what is a good idea? A good idea solves a problem that exists for a significant number of people, with scalability, differentiation, and understanding where the pain comes from. Is the pain coming from the absence of solution, from existing solutions, or from integration cost timing? Good entrepreneurs have a deep understanding of these things, the problems. I’ve read a lot of books, talked to many successful entrepreneurs, and those who spent more time understanding the pain point they were trying to solve rather than just designing a solution from scratch were the most successful. That’s a great insight.
Kieran Chandler: So, maybe this is our last question. Joannes, as someone who’s been there and done it, what advice would you give to someone who’s starting out with a startup?
Joannes Vermorel: I don’t know if I can really give advice. Lokad has been moderately successful, but we’re not Google yet. But I mean, my specific approach was to look for ancient, unsolved problems. When I started Lokad, it was with supply chain problems that were unsolved, but also very fundamental and basic.
Kieran Chandler: Just deciding how much to produce, where to produce, where to pile up your inventory is something very basic. I mean, even if we are transitioning toward a relatively digital economy where digital assets have a lot of importance, people still need to eat. So basically, there is like physical stuff that needs to be moved around. If it’s perishable and you stock too much, you will have expiration date issues. You will have to discard your inventory. And because the world has gone global, the downside of that is that the supply chains have become incredibly complex.
Joannes Vermorel: Absolutely. If you decide to produce every single consumer electronic device in like 20 different countries, it’s going to be complex. And there are a lot of inefficiencies. My focus was to identify problems that were relatively fundamental, that do not change that much. The physics of 3D printing is great, but it’s still not there yet. We are still not 3D printing entire cars. It works in B2B and industry, but it’s not working very well in the consumer space.
Kieran Chandler: As it stands today, 3D printing still is not very competitive.
Joannes Vermorel: Correct. So, the bottom line is, I try to identify relatively fundamental problems that will not change so much. Maybe the solution to the problem will change because you have some waves of new AI theories that would challenge how you can tackle the problem. But first, I thought, let’s identify a problem that itself is relatively stable. If you concentrate and repeat your efforts, you have a chance of having a problem that would not elude your grasp just because the problem has disappeared.
Kieran Chandler: What would be the opposite of your approach?
Joannes Vermorel: The opposite would be companies who are trying to do a Twitter app, which was the absolute opposite of what I was trying to solve. But again, that’s a matter of taste. It feels to me that the basic, uncool problems are still relatively undervalued in the startup world. For example, there are a lot of startups trying to do lifestyle products, but not many improving our garbage collection cycle or waste processing. Yet for the health of the world economy, disposing of the waste in a way that is safe, health-friendly and environment-friendly is a huge thing.
Olivier Ezratty: I agree, in the B2B space or the enterprise space, creating a product is the toughest thing to do. This is a skill that is not so broadly taught and known. It’s complicated to create products. Many startups think they create a product, but in the end, they offer a service. They have consultants and work on a project base for each customer. So, the big challenge and the discipline to understand is that creating a product is a discipline that’s mixing an understanding of your customer, marketing and business, and understanding of the technology you use to create the product.
Kieran Chandler: Why are there very few companies that do that well?
Olivier Ezratty: Some of the reason is it’s hard to fund because you need significant funding to pay for creating a product where you don’t get revenue for a while. And then maybe after one or three years, you get some revenue. If you don’t have enough revenue, you sell an unfinished product, and you need more service to sell it to your first customers. But then you become a service company. So there’s a kind of connection between the way you can raise enough money, maybe outside of your own country if you want to scale, like getting money from the US, and the way you create a product.
Kieran Chandler: Your work sounds very similar to what we do here. But I’m afraid we’re going to have to wrap it up today. Thanks very much for taking the time out to talk to us today.
Joannes Vermorel and Olivier Ezratty: Thank you.
Kieran Chandler: Okay, so that’s all for this week’s episode. We’ll be back again next week. Until then, bye for now.