483 startup failure post-mortems
Authors: CB Insights Tags: startups, venture capital, technology, business strategy, failure analysis Publication Year: 2024
Overview
In this report, we compile and analyze the post-mortems of 483 startups, offering a direct look into why promising companies fail. The core lesson is that failure is rarely due to a single cause; it is almost always multifactorial. While challenging macroeconomic conditions, such as the dramatic drop in global funding since 2021, create a difficult environment for everyone, they are often just one piece of the puzzle. Our analysis of founders’ goodbye letters, investor takedowns, and investigative journalism reveals a host of internal and external pressures that lead to a company’s demise. These include flawed business models, an inability to find [[product-market fit]], intense competition, management missteps, and sometimes, outright fraud. This compilation is intended for founders, investors, product engineers, and anyone in the tech ecosystem who wants to learn from the mistakes of others. By understanding the common pitfalls—from premature scaling to failing to adapt to evolving consumer interests—you can better navigate the turbulent waters of the startup world. The insights here are not just historical case studies; they are cautionary tales with direct relevance to current challenges in AI, fintech, biotech, and other cutting-edge sectors. We believe that a thorough understanding of failure is a prerequisite for building resilient, successful companies.
Book Distillation
1. 2024 First Update: Olive
Despite raising $850M and reaching a $4B valuation, this healthcare AI solutions company failed due to a combination of challenging economic conditions, evolving customer expectations, and critical management missteps. The company was ultimately forced to sell off its core business units and shut down its broader operations in October 2023.
Key Quote/Concept:
[[Misalignment of Product, Market, and Management]]: Olive’s failure highlights a classic trifecta of issues. Even with significant funding and a promising AI-driven product, a company can collapse if it fails to adapt to market shifts, meet customer needs, and maintain strong leadership.
2. 2024 First Update: Convoy
Convoy, the ‘Uber for trucking,’ struggled to convert its intentionally money-losing model into a profitable business. This fundamental business model flaw was exacerbated by a downturn in shipping demand and a contraction in the VC market, leaving it unable to raise further capital and leading to its closure in October 2023.
Key Quote/Concept:
[[Unsustainable Business Model]]: Convoy’s model, which relied on continuous cash infusions to operate at a loss, proved unsustainable when market conditions tightened. It serves as a warning against growth-at-all-costs strategies that lack a clear path to profitability.
3. 2024 First Update: Hyperloop One
The high-speed transportation startup failed to overcome immense technical and regulatory obstacles. Despite raising $472M and pivoting from passenger to cargo transport, it could not secure a single contract for a working system, leading to its shutdown and asset liquidation in December 2023.
Key Quote/Concept:
[[Technical and Regulatory Hurdles]]: Hyperloop One’s ambitious vision collided with the reality of building revolutionary infrastructure. The failure underscores the immense challenge of commercializing ‘moonshot’ projects that require overcoming both unsolved engineering problems and complex governmental regulations.
4. 2024 First Update: InVision
A once-dominant collaborative product design platform, InVision failed to innovate and allowed its products to grow stale. This created an opening for more agile competitors like Figma to capture the market, leading InVision to discontinue its core services by the end of 2024.
Key Quote/Concept:
[[Product Stagnation]]: InVision’s demise is a stark reminder that market leadership is temporary. Failing to continuously invest in product development and respond to competitive threats can lead to rapid decline, even for a well-funded company with a strong initial market position.
5. 2024 First Update: Ghost Autonomy
This autonomous driving software developer struggled to commercialize its technology despite multiple pivots and substantial funding, including from the OpenAI Startup Fund. The company couldn’t find a viable path to market for its consumer kits, crash prevention tech, or LLM-based driving solutions, ultimately shutting down in April 2024.
Key Quote/Concept:
[[Failure to Commercialize]]: Ghost Autonomy’s story illustrates that even with cutting-edge technology and strong backing, a startup can fail if it cannot translate its innovations into a marketable product that solves a clear customer need.
6. 2024 First Update: Zeus Living
Zeus Living, a real estate tech company subletting furnished properties for extended corporate stays, was severely impacted by rising interest rates. The increased cost of home buying prevented the company from generating sufficient revenue on its property investments, leading to its shutdown in November 2023.
Key Quote/Concept:
[[Macroeconomic Sensitivity]]: The failure of Zeus Living demonstrates how business models heavily reliant on specific market conditions, such as low interest rates in the real estate sector, are highly vulnerable to macroeconomic shifts.
7. 2023 Q3’23 Update: Koyo
Digital lending startup Koyo, which used open banking for consumer loans, failed after being unable to secure fresh equity funding amid broader economic turbulence. Despite having ample debt financing to issue loans, the lack of equity capital to fund operations forced it to wind down in July 2023.
Key Quote/Concept:
[[Capital Structure Imbalance]]: Koyo’s failure shows the critical difference between debt and equity financing. A business can have access to capital for its core lending function but still fail if it cannot secure the equity funding needed for operational runway and growth.
8. 2023 Q3’23 Update: Multichain
The cross-chain crypto protocol collapsed due to a single point of failure: its CEO. After the CEO was taken into custody by Chinese police, the team lost access to all operational funds and servers, which were solely under his control. This catastrophic operational risk forced the company to shutter.
Key Quote/Concept:
[[Operational Centralization Risk]]: Multichain’s collapse is a dramatic example of the dangers of centralizing critical operational control. The lack of distributed access and a clear succession plan created a fatal vulnerability that destroyed a company with a $5B Total Value Locked.
9. 2023 Q2’23 Update: Zume
The Softbank-backed robot pizza startup burned through nearly $500M trying to solve complex technological and logistical problems, such as preventing cheese from sliding off pizzas cooked in moving trucks. After failing to generate meaningful revenue, it pivoted to sustainable packaging but could not raise more funds and closed in June 2023.
Key Quote/Concept:
[[Over-engineering a Solution]]: Zume’s attempt to automate pizza delivery from moving trucks was a technologically complex solution to a problem that didn’t necessarily require it. The failure highlights the risk of focusing on ambitious engineering over a viable, revenue-generating business model.
10. 2023 Q2’23 Update: IRL
The social app IRL, which had achieved unicorn status, shut down after an internal investigation revealed that 95% of its 20 million active users were fake. The massive user fraud led to an SEC probe and the company’s complete collapse.
Key Quote/Concept:
[[User Metrics Fraud]]: IRL is an extreme cautionary tale about the pressure to show growth. Fabricating user metrics not only constitutes fraud but also makes it impossible to build a real business, as there is no genuine user base to monetize or serve.
11. 2023 Q1’23 Update: Argo AI
Despite a $1B commitment from Ford and a $7.3B valuation, the autonomous vehicle startup failed to secure new investors amid slow progress toward commercialization. Its main backers, Ford and VW, shifted their strategic focus away from the startup’s approach, leading to its shutdown.
Key Quote/Concept:
[[Loss of Strategic Backing]]: Argo AI’s failure demonstrates the vulnerability of startups that rely heavily on corporate partners. When those partners’ strategic priorities change, it can abruptly cut off the funding and support necessary for survival, especially in capital-intensive fields like autonomous driving.
12. 2022 First Update (1/11/23): Kite
AI-assisted coding startup Kite shut down after failing to find product-market fit. The company realized its product, which increased developer speed by 18%, was not compelling enough for engineering managers to purchase. Hitting the market too early and being unable to monetize led to its closure.
Key Quote/Concept:
[[Product-Market Misfit]]: Kite’s product was a ‘nice-to-have’ rather than a ‘must-have.’ It offered an incremental improvement that didn’t solve a significant enough pain point to justify its cost, illustrating the critical importance of aligning a product’s value proposition with a strong market need.
Generated using Google GenAI
Essential Questions
1. What are the primary, recurring reasons that well-funded startups fail, according to this analysis of post-mortems?
The report’s central argument is that startup failure is almost always multifactorial, not attributable to a single cause. The most prominent reason is the failure to find [[product-market fit]], as seen with the AI coding assistant Kite, which built a ‘nice-to-have’ tool that engineering managers wouldn’t pay for. Another critical factor is a flawed or [[unsustainable business model]], exemplified by Convoy, the ‘Uber for trucking,’ which relied on a continuous burn of venture capital without a clear path to profitability. Management missteps and an inability to adapt are also key; InVision, a once-dominant design tool, suffered from [[product stagnation]] and was overtaken by the more innovative Figma. Furthermore, some failures are due to catastrophic operational risks, such as Multichain’s collapse due to [[operational centralization risk]] where the CEO was a single point of failure, or outright fraud, as with the social app IRL, which faked 95% of its users. While macroeconomic headwinds create a difficult environment, these internal strategic and operational flaws are the ultimate drivers of failure.
2. How do macroeconomic conditions interact with internal company weaknesses to cause failure?
The report posits that challenging macroeconomic conditions, such as rising interest rates or a contraction in venture funding, act as catalysts that expose and exacerbate pre-existing internal weaknesses. They are rarely the sole cause of failure but rather the stress test that breaks a fragile system. For example, Zeus Living, a real estate tech company, was highly sensitive to rising interest rates, which made its model of subletting properties unprofitable. The macroeconomic shift didn’t create the business model’s vulnerability; it exploited it. Similarly, Convoy’s money-losing model was sustainable only as long as cheap venture capital was abundant. When the VC market tightened, its fundamental flaw was laid bare, and it could not raise the necessary funds to survive. For companies like Koyo, a digital lender, the inability to secure fresh equity funding for operations, despite having debt financing for loans, shows how a capital-constrained environment can starve a company of its operational runway. In essence, a market downturn removes the margin for error, making it impossible for startups with flawed models or weak market positions to continue.
3. What role does product innovation and strategy play in a startup’s survival, even after achieving market leadership?
The report underscores that product innovation is not a one-time event but a continuous necessity for survival. The case of InVision is a stark warning against [[product stagnation]]. Despite being a market leader in collaborative design with a $1.9B valuation, it allowed its products to ‘grow stale.’ This failure to innovate created a critical opening for a more agile and user-centric competitor, Figma, to capture the market, ultimately leading to InVision discontinuing its core services. This demonstrates that market leadership and significant funding provide no long-term protection without a relentless focus on product development and adaptation to evolving user needs and competitive threats. Conversely, Ghost Autonomy illustrates the challenge of [[failure to commercialize]] innovation. Despite multiple pivots and cutting-edge technology in the autonomous driving space, it couldn’t translate its technical prowess into a marketable product that solved a clear customer problem, proving that innovation without a viable commercialization strategy is insufficient for success.
Key Takeaways
1. Failure is Multifactorial, Not a Monocausal Event
The core lesson from analyzing 483 post-mortems is that startups rarely die from a single, isolated cause. While a founder might blame a tough funding environment, the reality is often a combination of factors: a weak business model, a product that doesn’t solve a burning need, management missteps, intense competition, and poor timing. For instance, the healthcare AI company Olive raised $850M but failed due to a ‘trifecta of issues’: challenging economic conditions, evolving customer expectations, and management missteps. This confluence of internal and external pressures creates a cascade of problems that becomes insurmountable. Understanding this complexity is crucial for avoiding simplistic diagnoses and building resilience by addressing weaknesses across all facets of the business, from product and market to finance and operations.
Practical Application: An AI product engineer should adopt a systems-thinking approach. Instead of focusing solely on the technical excellence of a model or feature, they must constantly ask how it fits into the larger business context. This means engaging with sales and marketing to understand customer needs ([[product-market fit]]), talking to finance about the cost implications of technical decisions ([[business model viability]]), and being aware of competitive and market shifts. For example, when building a new AI feature, they should validate not just its accuracy but also its business value, its defensibility against competitors, and its alignment with the company’s path to profitability.
2. A Sustainable Business Model is More Important Than Growth-at-all-Costs
The era of celebrating growth while ignoring profitability is over, as demonstrated by multiple failures in the report. Convoy, the ‘Uber for trucking,’ is a prime example. Its model was intentionally money-losing, predicated on the assumption that it could always raise more venture capital to subsidize operations and capture market share. When the VC market contracted, this [[unsustainable business model]] collapsed. This serves as a powerful cautionary tale against strategies that prioritize burning cash for growth without a clear and plausible path to positive unit economics and eventual profitability. The market now demands not just innovation and user acquisition, but a viable plan to build a self-sustaining enterprise. Startups that ignore this fundamental principle are building on a foundation of sand, vulnerable to the first significant market downturn.
Practical Application: When designing an AI product, an engineer must consider monetization from day one. This doesn’t mean sacrificing user experience for short-term revenue, but it does mean building a product where the value created is clear enough that customers will eventually pay for it. For instance, instead of offering a powerful AI tool for free indefinitely to maximize user count, an engineer could help design a freemium model where the core value is free but advanced, high-resource features (e.g., complex model training, detailed analytics) are paid. This ensures the product strategy is aligned with long-term financial sustainability, not just vanity metrics.
3. Product Stagnation is a Death Sentence in a Competitive Market
The demise of InVision, once a dominant force in product design valued at $1.9B, is a stark reminder that market leadership is perishable. The report highlights that users felt the company ‘allowed its products to grow stale,’ which created the opportunity for Figma to rise. InVision’s failure to continuously invest in its core product, listen to its users, and respond to competitive threats led directly to its decline. This illustrates a critical principle: a startup’s product is not a static asset but a dynamic entity that must constantly evolve. Complacency, technical debt, and a loss of focus on the user experience can quickly erode even the strongest market position, as more agile and innovative competitors will inevitably emerge to fill the void.
Practical Application: An AI product engineer must champion a culture of continuous improvement and iteration. This involves setting up robust feedback loops with users to understand their evolving pain points, actively monitoring the competitive landscape for new technologies and approaches, and advocating for resources to address technical debt and refactor systems before they become a drag on innovation. For example, they could implement an A/B testing framework to experiment with new AI models or UI changes, ensuring that product decisions are data-driven and user-validated, thus preventing the product from becoming stale and irrelevant.
4. Beware of Catastrophic Single Points of Failure
Some failures are not slow burns but sudden collapses resulting from a critical, centralized vulnerability. The story of Multichain is the most dramatic example. The entire crypto protocol, with $5B in Total Value Locked, was destroyed because its CEO was the only person with access to operational funds and servers. His arrest by Chinese police was a fatal, single point of failure. This highlights the extreme danger of [[operational centralization risk]]. While less dramatic, the failure of Argo AI also shows a dependency risk. Its survival was contingent on the continued strategic backing of Ford and VW. When their priorities shifted, Argo’s support vanished. These cases demonstrate the need to de-risk operations by distributing control, knowledge, and dependencies.
Practical Application: An AI product engineer should advocate for robust and decentralized engineering and operational practices. This includes meticulous documentation so that knowledge isn’t siloed with one person, implementing infrastructure-as-code to ensure systems are reproducible, and establishing clear on-call and incident response protocols that don’t rely on a single ‘hero’ engineer. In a machine learning context, this means versioning datasets and models (e.g., using DVC, MLflow) and automating training and deployment pipelines so that the process is not dependent on one person’s laptop or manual steps. This builds a resilient system that can withstand unexpected personnel changes or other disruptions.
Suggested Deep Dive
Chapter: 2023 Q2’23 Update: Zume & IRL
Reason: This section provides two of the most vivid and cautionary tales in the entire report. Zume’s story of burning nearly $500M to solve the problem of cheese sliding off pizzas in moving trucks is a perfect case study in [[over-engineering a solution]] and losing sight of business fundamentals. It’s a crucial lesson for engineers who can fall in love with a technical challenge without asking if it’s the right problem to solve. Immediately following it, the story of IRL, which fabricated 95% of its users, is an essential ethical case study on the pressures of the ‘growth-at-all-costs’ mindset and the ultimate futility of [[user metrics fraud]]. Reading these two back-to-back provides a powerful one-two punch on the importance of building a real business for real customers.
Key Vignette
The Robot Pizza Truck That Couldn’t Keep the Cheese On
Zume, a Softbank-backed startup, raised nearly half a billion dollars with the ambitious vision of disrupting the pizza industry with robotic automation. The plan was to equip delivery trucks with robotic pizza-makers and smart ovens to cook pizzas en route to customers. However, the company became obsessed with solving complex technological problems, such as how to prevent melted cheese from sliding off the pizzas while the trucks were in motion. This focus on an [[over-engineered solution]] for a minor problem, combined with a high cash burn rate, meant the company failed to generate meaningful revenue and eventually collapsed.
Memorable Quotes
‘Since the inception of the project, all operational funds and investments from investors have been under Zhaojun’s control. This also means that all the team’s funds and access to the servers are with Zhaojun and the police,’ the tweet said…
— Page 20, 2023 Q3’23 Update: Multichain
While Koyo was flush with debt funding to ramp up its lending—a chunky £100m raised from Atalaya Capital Management just nine months ago—difficulties securing additional equity funding forced the fintech to appoint advisers and eventually trigger a wind down.
— Page 17, 2023 Q3’23 Update: Koyo
We were operating under the assumption that we just needed one last funding round that would then get us to profitability; and this wasn’t just a pipe dream… the cold truth was we were up against an investment drought with a model no longer favored by VC investors to produce venture returns.
— Page 22, 2023 Q2’23 Update: Casai
…Zume had struggled with problems like stopping melted cheese from sliding off its pizzas while they cooked in moving trucks, per Bloomberg. Its difficulties led to a string of high-profile departures and financial problems.
— Page 29, 2023 Q2’23 Update: Zume
Ford said in its third-quarter earnings report released Wednesday that it made a strategic decision to shift its resources to developing advanced driver assistance systems, and not autonomous vehicle technology that can be applied to robotaxis…
— Page 53, 2022 First Update (1/11/23): Argo AI
Comparative Analysis
This CB Insights report serves as an empirical, evidence-based counterpoint to the prescriptive frameworks found in canonical startup literature like Eric Ries’s ‘The Lean Startup’ or Peter Thiel’s ‘Zero to One.’ While Ries provides the ‘how-to’ for building a company through validated learning and the Build-Measure-Learn loop, this report provides 483 case studies of ‘what happens when you don’t.’ The failure of Kite to find [[product-market fit]] is a real-world manifestation of failing the core tenets of the lean methodology. Similarly, where Thiel argues for creating a monopoly through unique technology and a strong business model, the post-mortems of companies like Hyperloop One (technical hurdles) and Convoy (unsustainable business model) illustrate the immense difficulty of achieving that vision. Unlike those books, which offer forward-looking strategies from a successful founder’s perspective, this report is a backward-looking analysis of failure from the perspective of journalists and the founders themselves. Its unique contribution is not a new theory, but a sobering, data-grounded catalog of common failure patterns, making it an essential companion piece that grounds the abstract principles of other works in the harsh reality of the startup graveyard.
Reflection
This compilation of post-mortems is an invaluable resource, acting as a sobering corrective to the often-hyped narrative of startup success. Its greatest strength lies in its breadth, aggregating hundreds of failures to reveal undeniable patterns. The recurring themes of failing to find [[product-market fit]], premature scaling, and flawed business models are not just theoretical concepts here; they are backed by the painful experiences of companies that burned through hundreds of millions of dollars. The authorial voice is that of a data-driven analyst, which lends credibility but can sometimes lack a deeper narrative synthesis. A potential weakness is the inherent bias in the source material; it only includes startups whose founders chose to write a public post-mortem, potentially omitting failures due to more sensitive or litigious reasons. Critically, while the report acknowledges macroeconomic pressures, its most vital insight is that these external factors typically only accelerate the demise of companies already suffering from fundamental internal weaknesses. For an AI product engineer, this is a crucial lesson: it is tempting to blame the market, but success and failure are most often determined by the relentless, unglamorous work of building a valuable product that customers will pay for, supported by a sustainable business model.
Flashcards
Card 1
Front: What was the primary reason for the failure of InVision, a once-dominant product design platform?
Back: [[Product Stagnation]]. The company failed to innovate and allowed its products to grow stale, which enabled more agile competitors like Figma to capture the market.
Card 2
Front: What catastrophic operational risk did the collapse of the crypto protocol Multichain exemplify?
Back: [[Operational Centralization Risk]]. The CEO was the single point of failure. His arrest by Chinese police resulted in the team losing access to all operational funds and servers, forcing the company to shutter.
Card 3
Front: Why is Convoy (‘Uber for trucking’) cited as a cautionary tale about business models?
Back: It had an [[Unsustainable Business Model]]. Its growth-at-all-costs strategy relied on continuous infusions of venture capital to operate at a loss, which proved fatal when the VC market contracted.
Card 4
Front: What critical lesson does the failure of the social app IRL, which claimed 20 million users, teach about metrics?
Back: The danger of [[User Metrics Fraud]]. An internal investigation revealed that 95% of its users were fake, making it impossible to build a real business and leading to an SEC probe and its collapse.
Card 5
Front: What is [[Product-Market Misfit]], as illustrated by the AI-assisted coding startup Kite?
Back: The product was a ‘nice-to-have’ rather than a ‘must-have.’ It offered an incremental improvement (18% faster coding) that wasn’t compelling enough for engineering managers to justify purchasing it.
Card 6
Front: What was the core strategic flaw of Zume, the Softbank-backed robot pizza startup?
Back: [[Over-engineering a solution]]. It burned through nearly $500M on extreme technological complexity (e.g., preventing cheese from sliding on pizzas in moving trucks) instead of focusing on a viable, revenue-generating business model.
Card 7
Front: How did corporate strategy lead to the shutdown of the well-funded autonomous vehicle startup Argo AI?
Back: [[Loss of Strategic Backing]]. Its primary corporate backers, Ford and VW, shifted their strategic focus away from Argo’s approach to autonomous driving, abruptly cutting off the capital and support necessary for its survival.
Generated using Google GenAI