Andrew Chen Essay Collection

Andrew Chen | The Cold Start Problem: How to Start and Scale Network Effects | Talks at Google

BRAINDUMP ON VIRAL LOOPS - by Andrew Chen - @andrewchen

Viral loops are systematically engineered product features designed to reach millions of users through measurable, repeatable mechanics. Unlike one-time marketing spikes, these loops are built into the product via invites, tagging, and referral links. The core metric is the viral factor, a ratio representing how many new users each cohort signs up. A viral factor of 1.0 indicates exponential growth, while anything below 1.0 eventually stops, though factors as low as 0.2 still provide significant customer acquisition cost discounts by providing free users. Products generally fall into two categories. Category 1 includes simple, highly shareable apps like generative AI tools or photo filters that grow rapidly but often face retention challenges. Category 2 consists of complex, sticky products like Slack or Figma that grow more slowly but maintain high long-term value. The transition from Web 2.0 to mobile fundamentally changed these mechanics. While email address book imports once allowed for massive invite volumes, mobile platforms restricted contact access to one-by-one selection, lowering the number of invites per session. Consequently, modern virality relies more on retention than initial session spamminess. High-retention products generate viral factor across multiple sessions rather than just the first. This cumulative effect means that even unobtrusive sharing features can lead to a viral factor greater than 1.0 if the user returns frequently. Current AI tools are revitalizing the create-and-share loop because their visual outputs are highly resonant with modern social media platforms. However, long-term success still requires transitioning from viral spikes to durable retention and network effects to avoid the inevitable decay caused by novelty effects and market saturation.

The Anti-Pitch: When haters hate your startup idea

The Anti-Pitch is a concise, skeptical one-liner designed to invalidate a startup idea by targeting perceived weaknesses. While the traditional elevator pitch focuses on potential, the Anti-Pitch leverages common industry memes to shut down conversation. For AI startups, this often manifests as the GPT wrapper critique. For consumer hardware, it centers on inventory risk and holiday cycles. For social apps, it focuses on the dominance of existing network effects. Business Pain / Root Cause: Founders often face reflexive skepticism that can derail fundraising or recruitment. The root cause is the high statistical failure rate of startups, which makes skepticism a safe default position for observers. Since 99 percent of startups fail, skeptics are statistically correct most of the time, making their rebuttals feel authoritative despite being intellectually lazy. Founders have three primary options when facing these rebuttals. First, they can ignore the critic if the person is not a key stakeholder, such as a journalist or a non-industry peer. Second, they can accept the criticism but address it with a specific innovation, such as being the first AI-native version of a legacy tool. Third, they can deny the association, though this risks reinforcing the negative comparison. The most effective long-term strategy is the Pitch Maze. This involves iterating the narrative to find a path that avoids predictable dead ends. For example, reframing a gambling app as a predictive markets product changes the nature of the skepticism and the regulatory conversation. While founders should listen for shreds of reality in criticism to avoid the show up and throw up sales trap, they must ultimately navigate the rhetorical landscape to protect their optimistic vision.

Why retention is so hard for new tech products

Retention is the primary indicator of product-market fit and follows predictable geometric decay patterns. Initial Day 1 (D1) retention typically dictates long-term viability. A 10% D1 retention rate is generally unfixable through marginal optimizations like notifications or A/B testing. Meaningful improvements require significant pivots in core product architecture or positioning. The business pain often stems from attempting to optimize a product that lacks fundamental utility for its target audience. A critical distinction exists between usage retention and revenue retention. In B2B SaaS, revenue often expands within a cohort even as the number of active users declines. This dynamic is driven by organic adoption and increased seat count. Consumer products rarely benefit from this, though platforms like Amazon or Uber achieve similar results by expanding the range of services used by a single customer. This makes B2B business models structurally more resilient than most consumer counterparts. Product categories impose natural limits on retention. A travel app will never achieve the daily usage frequency of a messaging platform or a coding tool. Success in high-frequency categories usually requires displacing an existing daily habit. This displacement is often achieved through a 20% remix of an existing category rather than total reinvention. The Golden Cohort consists of early, high-intent users who find the product organically. As acquisition scales to broader markets, retention metrics typically degrade. Churn is fundamentally asymmetric: losing a user is effortless, while resurrecting a dormant user is prohibitively expensive and usually requires network-driven reactivation rather than marketing prompts. True magic retention stems from fresh market insights or technological shifts, such as LLMs or societal changes. These Why Now factors allow a product to solve a known problem in a viscerally better way within the first 60 seconds of use. Timing and category selection are as critical as the product itself.

AI will change how we build startups -- but how?

The current state of AI adoption reveals a significant gap between technological potential and daily utility, as evidenced by the Home Screen Test. Most mobile home screens remain devoid of AI-native applications, suggesting we are in the earliest stages of the transition. The primary shift will not just be in the products themselves but in the fundamental structures used to build them. Key uncertainties include whether AI leverage will lead to billion-dollar one-person companies or if human bottlenecks like taste and design will necessitate large teams. Defensibility is a major concern; as AI commoditizes software creation, moats may shift toward network effects, brand, or high-CapEx industries like space tech and hardware where AI cannot easily arbitrage value. The cost of building may decrease, but the cost of distribution and user acquisition remains a significant barrier. Organizational roles in engineering, product, and design are likely to collapse into multimodal disciplines, potentially giving rise to the Product Engineer who manages features end-to-end. This shift could further decentralize tech hubs like San Francisco, as building products becomes more akin to content creation that can happen anywhere. Venture capital may also evolve, moving from risk-based funding for experimentation to growth-based funding for profitable, fragmented startups that jump directly from zero to Series A.

Lies per Second, Meetings per Decision Ratio, and other important biz metrics

Andrew Chen introduces the concept of micromorts, a unit representing a one in a million chance of death, to illustrate how risk and intangible experiences can be quantified. This framework extends into personal life through metrics like Cost per Hour of Pleasure (CPHP) and Phone Pickups per Hour, which measure the quality of time spent. The primary focus shifts to unconventional business metrics designed to diagnose startup health and operational efficiency. Lies per Second (LPS) assesses the integrity of a pitch, while the Meetings per Decision Ratio (MPDR) highlights organizational bottlenecks where authority is unclear or processes are ill-defined. Time to First Excuse (TFE) tracks team accountability, suggesting that a low TFE often precedes the need for a leadership reboot. The Numbers vs Text Ratio serves as a maturity indicator for startups moving toward Series B, where narrative-heavy decks should be replaced by dense financial and retention data. PowerPoints per Launch (PPPL) measures the creep of bureaucracy, signaling when a company has shifted from a shipping culture to one of consensus-building hell. Finally, the Decision to Rumination Ratio explores the paradox of spending weeks on minor purchases while making major life decisions impulsively. Chen posits that future AI tools will likely automate the tracking of these metrics by analyzing meeting transcripts and fact-checking presentations in real-time.

Corpospeak: Why you still sound like a faceless corporate entity

Traditional marketing is undergoing a massive reordering as centralized mass communication channels dissolve into millions of microniche channels. This shift renders the traditional corpospeak playbook, characterized by formal, trite, and faceless brand communication, ineffective. The reliance on polished, sound-bite-driven messaging often pushes customers away in an era where authenticity and vibes dictate engagement. The business pain stems from a reliance on outdated playbooks that prioritize brand polish over real connection, leading to diminishing returns on traditional CAC and SEO investments. The root cause of this failure is the functional separation between marketing and R&D. When professional marketers or PR agencies handle communication, the resulting output is often wooden and lacks the substantive depth customers crave. This PR/brand industrial complex prioritizes polish over reality, creating a DMV-like experience for customers who are forced to interact with support bots or junior staff instead of the builders who understand the product. This gap between the creator and the communicator is the fundamental reason why corporate entities sound faceless. Delegation is no longer a viable strategy for social media or community engagement. Outsourcing these functions to social media interns or agencies fails because customers want to hear from founders, engineers, and designers. These builders possess the passion and technical knowledge that resonates with hardcore users. The new normal requires a decentralized approach where every employee shares the responsibility of talking to customers online. Success in this environment favors quantity over quality and brand ubiquity over brand dilution. Founders and CEOs must lead this transition by posting from personal accounts, nerding out on technical details, and communicating their personal motivations. Rather than protecting against downside risk through legal and compliance gatekeepers, companies must build frameworks that enable direct, frequent, and authentic interactions.

Every marketing channel sucks right now - by Andrew Chen

Marketing channels are currently in a state of terminal maturity, characterized by declining engagement and prohibitive costs. Established Big Channels such as SEO, paid advertising, and influencer marketing suffer from the Law of Shitty Clickthroughs: a phenomenon where marketing strategies inevitably decay as competition increases and consumer novelty wears off. For early-stage startups, these mature channels often represent a trap because they lack the lifetime value (LTV) and brand equity of incumbents to absorb low ROI. SEO is increasingly dominated by listicles and Reddit, while influencer marketing often results in traffic spikes without conversion. Paid ads have become an auction-driven race to the bottom where competitors easily copy creative assets. The current landscape suggests a shift toward Little Channels: unscalable, manual tactics like niche community engagement, localized events, and direct outreach. These methods succeed by leveraging novelty and high-touch interaction that larger competitors cannot replicate. Startups should prioritize these asymmetric tactics, even if they are one-time events, to build initial momentum. Furthermore, marketing serves as a multiplier for product quality. In the current AI-driven tech cycle, viral growth is often a byproduct of high product novelty rather than sophisticated funnel optimization. However, as AI outputs become commoditized, the novelty effect will fade, necessitating a return to innovative, perhaps polarizing, brand strategies. Startups are encouraged to embrace unscalable tactics to prove product-market fit before attempting to solve for repeatability in mature channels.

Why a16z is investing up to $1M in very early stage startups

Andrew Chen, General Partner at Andreessen Horowitz, details the a16z Speedrun program, a specialized accelerator targeting very early-stage startups in tech, entertainment, and AI. The program offers up to $1M in direct investment and $5M in credits from major infrastructure partners including AWS, GCP, OpenAI, Microsoft, NVIDIA, Stripe, and Deel. Participants undergo a 90-day intensive cycle featuring mentorship from founders of high-growth companies like Figma, Zynga, and DoorDash. Chen notes a significant shift in the venture landscape, observing that the AI revolution has re-centered startup activity in San Francisco despite post-pandemic geographic dispersion. Speedrun aims to bridge the gap for founders who are too early for traditional $10M to $20M a16z checks but require the firm's extensive operational resources in hiring, marketing, and financing. The fifth cycle of the program is scheduled for July 28 to October 10, 2025, with an application deadline of May 11. The initiative reflects a strategic move to capture value from first-time and repeat founders at the messy idea stage before they reach significant revenue milestones. This approach leverages the firm's 600-person team to provide hands-on support in hiring, partnerships, and future financing rounds.

DoorDash's v1 was 8 PDF menus on a static HTML website

DoorDash began as Palo Alto Deliveries, a static HTML site featuring eight PDF menus and a Google Voice number. This 45-minute build serves as a classic example of a Minimum Viable Product (MVP). However, the traditional MVP framework popularized by The Lean Startup faces significant challenges in the current technology landscape. Modern product development often suffers from inconclusive testing and false negatives. When an MVP is too minimal, it may fail not because the core idea is flawed, but because it lacks the polish or features required to compete in mature markets. This leads teams to abandon promising directions prematurely. Business Pain / Root Cause: The primary pain point is the looping treadmill of inconclusive experiments. The root cause is often an over-reliance on easily measurable data rather than qualitative customer insight. In mature categories like design tools or productivity software, users expect high levels of polish from day one. Furthermore, emerging sectors like AI foundation models or deep tech require massive upfront R&D, making a truly minimal version impossible. Success now requires building encyclopedic market knowledge and exercising strategic judgment rather than relying solely on iterative A/B testing. The startup journey is not about a single launch but a long series of potential failures punctuated by glimmers of interest. Reducing a vision to a single minimal version only addresses the first few steps of a much longer journey.

Vibe coding, some thoughts and predictions - by Andrew Chen

Vibe coding represents a fundamental shift in software development where natural language prompts replace manual coding. Coined by Andrej Karpathy, the term describes a workflow where creators rely entirely on LLMs to generate, debug, and iterate on code. This process often involves "accepting all" AI suggestions without reading diffs, allowing the code to grow beyond the creator's immediate comprehension while remaining functional. Tools like Cursor, Replit Agent, and models such as Claude 3.7 Sonnet and Grok 3 are driving this trend. The democratization of code will likely lead to software being dominated by youth culture, similar to the evolution of social media. As the cost of creation drops, software will become a disposable commodity. This shift may reduce the need for traditional open-source libraries because AI can generate custom code from scratch for every specific use case. The current command-line interface style of prompting is expected to evolve into "vibe designing," a visual paradigm where creators show the AI desired outcomes rather than typing instructions. Strategic bottlenecks in this new era will shift from technical execution to distribution, network effects, and consistent creativity. Adaptive software will emerge, where products automatically reconfigure their own UI and logic based on user behavior to achieve specific business outcomes. This will accelerate the penetration of software into long-tail industries that were previously too small or "boring" for high-end engineering teams to target. Organizational structures will also face disruption. The traditional ratio of engineers to designers and product managers will change. While engineering costs may decrease, Jevon's Paradox suggests that the demand for builders might actually increase as the speed and volume of software production accelerate. However, managing edge cases, security, and complex business logic remains a significant challenge for purely vibe-coded projects.

The 30 min interview for people with busy calendars

Where will the AI Horde strike next? AI video, social media, and Hollywood

The entertainment industry faces a systemic shift as the AI Horde, a collective of thousands of venture-backed startups, leverages generative video, speech, and music models to challenge legacy incumbents. Unlike traditional management theory which focuses on incumbent defense, the startup perspective is characterized by probabilistic exploration and offensive maneuvers targeting market weaknesses. **Business Pain / Root Cause:** Legacy media incumbents suffer from broken business models, high production costs, and rigid creative processes. While business leaders recognize AI's potential to increase quantity and reduce costs, they face significant friction from unions and a culture of technology aversion. This structural resistance creates a 'lag' that allows the AI Horde to capture market share in more flexible, digital-native environments. **Strategic Entry Paths:** Startups are currently navigating three primary paths. First, working with incumbents through the 'Pixar for AI' model or selling B2B tools for dubbing and editing. Second, building for digital-native platforms like YouTube and TikTok, which includes creating AI-generated media networks or creator-centric 'picks and shovels' tools. Third, developing standalone apps that target specific formats like micro-narratives or real-time interactive experiences. **AI-Native Innovation:** The most significant opportunity lies in real-time interactive content. By collapsing the traditional loop between production and consumption, AI enables a new genre where stories and environments react to user input in real-time. This shifts the value proposition from static viewing to dynamic engagement, effectively merging the mechanics of gaming with the narrative depth of film. Ultimately, the AI Horde moves as a wave, reinventing Hollywood, gaming, and social media simultaneously by filling every available niche.

The Growth Maze vs The Idea Maze - by Andrew Chen

AI products are entering a phase shift where the focus is moving from technical infrastructure to distribution. The initial AI boom was defined by solving the Idea Maze: a concept from Balaji Srinivasan describing the complex decision tree of product features, market history, and avoiding past failures. As AI becomes a general-purpose technology, technical defensibility is eroding due to fast-following and copycatting. The competitive landscape is now shifting toward the Growth Maze, which represents the path-dependent decisions required to reach mainstream markets. Traversing the Growth Maze involves selecting initial target audiences, choosing between bottoms-up PLG or enterprise sales, and determining the timing for scaling initiatives. Unlike the Idea Maze, which is navigated by iterating on features, the Growth Maze is solved through GTM experimentation and strategic hiring. In the AI era, this includes leveraging agentic sales tools and AI-driven marketing orchestration. The role of the marketer is evolving from campaign execution to high-level strategy, overseeing agents that generate thousands of creative variations and target hyper-segmented audiences. Historically, technology cycles begin with a focus on the Idea Maze when the novelty of a new tool is enough to win. As the S-curve matures, distribution becomes the primary differentiator. AI is reaching this inflection point where superior distribution coupled with a good enough product will likely outperform technically superior products with poor GTM execution. The root cause of this shift is the commoditization of AI building blocks, which forces founders to master distribution to maintain a competitive advantage.

Revenge of the GPT Wrappers: Defensibility in a world of commoditized AI models

The AI landscape is shifting away from foundation model dominance toward application-layer defensibility. Initial theories suggested that massive compute and data requirements would create insurmountable moats for model providers. However, the rapid rise of open source alternatives like DeepSeek and Grok, combined with model distillation, has compressed the lead time for state-of-the-art models to roughly six months. This commoditization suggests that the underlying model is no longer a sustainable competitive advantage. Instead, the industry is entering a phase reminiscent of the Web 2.0 transition. During that era, expensive proprietary stacks were replaced by open source infrastructure and database wrappers or CRUD apps. The winners were not those with the best database technology but those who built robust network effects and distribution channels. For modern AI startups, this means moving beyond the novelty effect of generative features to focus on three pillars of network effects. First, acquisition network effects leverage AI-generated content to drive organic sharing. Second, retention effects utilize multiplayer workflows, such as tagging and collaborative editing, to keep users engaged. Third, monetization effects capitalize on team-wide adoption and enterprise-grade features. B2B SaaS companies will likely find defensibility through deep workflow integration, compliance, and security rather than technical model superiority. A critical distinction in this cycle is that AI has not yet introduced a new user experience modality. Because users still interact via mobile and web interfaces, incumbents with existing distribution networks, such as WhatsApp or Salesforce, maintain a significant advantage. Startups must decide whether to build novel network features or risk being absorbed by incumbents who integrate AI into their established ecosystems. The speed of this cycle is significantly faster than previous computing shifts, requiring a rapid transition from product novelty to structural defensibility.

My top essays from 2024 - by Andrew Chen - @andrewchen

Andrew Chen's 2024 retrospective synthesizes critical shifts in the technology landscape, focusing on the intersection of AI, consumer behavior, and growth methodology. A primary theme is the emergence of Dopamine Culture, where AI applications experience rapid initial adoption driven by novelty but suffer from poor long-term retention as consumer patience thins. This phenomenon necessitates a shift in product strategy toward sustainable engagement rather than temporary social media spikes. Chen also signals the end of the era for billion-user, ad-supported consumer startups, suggesting that the next wave of value will come from highly-monetizing, utility-focused vertical applications. In the realm of organizational productivity, the distinction between 10x work and 1x work is highlighted, advocating for the elimination of thoughtless daily routines and endless meetings in favor of high-impact strategic initiatives. The evolution of growth hacking over the last decade is reviewed, noting how the discipline has matured beyond simple tactics into a complex integration of data and product design. Chen also explores the Bureaucrat mode trap, where excessive collaboration and consensus-seeking stifle innovation and speed. He argues that the road to organizational failure is often paved with stability and inclusiveness at the expense of momentum. Marketing is identified as a field on the verge of total reinvention due to AI. In a world characterized by infinite labor and content, mass personalization becomes the baseline, fundamentally altering how brands interact with users. Furthermore, the document addresses the limitations of data-driven decision-making, acknowledging that while data is essential, human intuition and the recognition of novelty effects are crucial when data becomes impossible to interpret clearly. The collection concludes with a call to Always Be Launching, emphasizing that consistent market presence outweighs the pursuit of a singular, perfect product debut. This strategy counters the reality that most new products are met with initial indifference, requiring persistent iteration to find traction.

My product went viral on social media but all I got were these shitty users

Viral social media spikes often result in an Invasion of Looky-Loos, defined as users who view a product with no genuine intention of purchasing or remaining active. While these spikes create impressive top-of-funnel metrics, they are typically ephemeral and non-durable. The core issue is the fundamental tradeoff between quality and quantity. Rapid user acquisition through viral loops generally leads to lower intent and poor retention. This phenomenon aligns with the Conservation of Intent principle: any mechanism that increases the top of the funnel by reducing friction inevitably lets in lower-intent users who are less motivated to complete the journey. To evaluate true traction, product builders should apply three specific lenses: durability, scalability, and value. Durable traction involves high-quality users who stay engaged over time. This is evidenced by cohort retention curves that flatten and consistent weekly growth between 5% and 10%. Scalable traction requires repeatable marketing motions in deep channels like referrals or paid marketing, rather than one-off viral hits that cannot be systematically replicated. Valuable traction focuses on high-intent users who convert to paid versions and drive key action metrics. Viral spikes can be counterproductive by overwhelming a product with segments it is not yet equipped to handle, such as international users or specific demographics. These users may use the product in unintended ways, potentially alienating the loyal core base. However, spikes can be leveraged through strategic friction, such as waitlists, to filter for quality. They can also serve as a signal to raise venture capital during a period of high visibility. Ultimately, sustainable product-market fit is defined by rigorous metrics: DAU/MAU over 50%, annual retention above 65%, and power user curves showing a smile concentration. The objective is scaling high-intent, sticky users rather than chasing ephemeral social media metrics that do not translate to long-term business value.

Why your product idea sounds too complicated

Product complexity is measured on a scale ranging from Simple to Double WTF, primarily defined by how a product is described to potential users. At the simplest level, a product is defined by an adjective plus noun construction, such as electric car or smart phone. This format leverages existing mental models while highlighting a singular, transformative change. As descriptions move down the scale, they become increasingly obscured by niche categorizations, weird behaviors, or word salad that focuses on technical mechanics rather than user value. The most complex descriptions often involve long-winded historical preambles or the misapplication of academic frameworks like jobs to be done before the actual product utility is even mentioned. Simplicity functions as a tangible competitive advantage. When a product is easy to describe, it spreads faster through organic word of mouth, onboarding becomes more efficient, and customer acquisition costs are reduced. To rectify a complicated description, builders must engage in the painful process of asking target customers to describe the product in their own words. This feedback often strips away the founder's perceived strategic differentiation and technical wow factor, but it provides an unvarnished truth that is essential for market resonance. Effective positioning often requires horseless carriage thinking, which bridges the gap between what customers understand and the innovation being delivered. If a pitch feels too bland, differentiation can be achieved through counterpositioning, acting as the anti-X to an established incumbent, or by making fundamental, visible choices in user experience, such as opting for a template-driven UI over a blank text box. Other levers include targeting specific audience segments, such as international markets or premium tiers, rather than relying on feature bloat. Ultimately, the ego of the builder often drives complexity, but market success depends on how clearly a product fits into a customer's existing life and specific use cases.

Always Be Launching - by Andrew Chen - @andrewchen

The traditional Big Bang Launch (BBL) model, characterized by a single spike of attention, is increasingly ineffective in a post-scarcity media landscape. In an era of infinite feeds and short news cycles, the Always Be Launching (ABL) philosophy advocates for a continuous narrative rather than a one-time event. This approach prioritizes a steady drumbeat of updates, including screenshots, demos, and case studies, to maintain market presence. ABL shifts the focus from advertising to relationship building, emphasizing communication over broadcasting. It encourages founders to go direct to their audience, bypassing traditional gatekeepers like journalists and PR agencies who often lack the distribution power of social platforms. The strategy relies on authenticity, where unvarnished insights and personal engagement from founder accounts drive higher engagement than corporate marketing. By publicizing small wins and lessons learned, companies accumulate reach over time. This cumulative effect is more sustainable than the episodic spikes of BBL, which are often forgotten within hours. The primary barrier to ABL is psychological, as founders may feel the need to be perfectly polished before speaking. However, lowering the bar for sharing metrics, bugs, and daily milestones builds a loyal community. Success in this model requires a shift from a scarcity mindset to an abundance mindset, where quantity and frequency are leveraged to build long-term brand equity and direct customer relationships.

Bureaucrat mode - by Andrew Chen - @andrewchen

Andrew Chen explores Bureaucrat Mode, the antithesis of Founder Mode, defining it as the state large organizations enter when collaboration and consensus-seeking override decisive action and conviction. While Founder Mode and Wartime CEOs emphasize high-conviction leadership, Bureaucrat Mode is the more common reality for scaled companies. Characteristics include committees for every decision, pre-meetings for meetings, lack of individual ownership, and the punishment of initiative. These behaviors often stem from good intentions like inclusiveness and stability, but when industrialized, they paralyze progress. Chen introduces the concept of self-replicating bureaucrats: individuals who thrive in process-heavy environments and hire others like them. In these systems, impact is often mismeasured by team size or project scope rather than actual output, creating incentives for hiring bloat and vanity projects. The document describes a cycle of life in tech where nimble startups win markets, hire managers to scale, eventually attract bureaucrats, lose their entrepreneurial talent, and ultimately become vulnerable to the next wave of startups. This cycle is essential for industry evolution but difficult to counteract, even in companies like Uber that prioritize speed. Highly consensus-driven organizations are structurally incapable of responding to unknowable shifts like AI or aggressive new market entrants because their processes require a level of certainty that does not exist in volatile environments.

Unfortunate startup pivots (a short list) - by Andrew Chen

Startups frequently pivot during early stages, but specific patterns of failure emerge when changes are reactive rather than strategic. Unfortunate pivots often involve moving from B2B to consumer markets. In these cases, founding teams often lack the necessary DNA for consumer engagement, even if they possess deep enterprise sales networks. Another common failure is the attempt to fix low retention by adding social features, chat, or notifications. These additions rarely succeed because the underlying product experience remains leaky. Indifferent users do not engage with secondary or tertiary features when the core value proposition is missing. Similarly, bolting on trendy technologies like AI or Web3 provides only a temporary novelty effect. These haphazard additions fail to engage users deeply and often appear superficial to both customers and investors compared to native solutions. The premature platform pivot occurs when a company attempts to support multiple verticals after failing to gain traction in a specific one. This zooming out strategy typically fails because it lacks a killer app or a demonstration of value in any single area. If a specific example does not work, a collection of non-working verticals will not improve the outcome. Conversely, zooming in on a specific, high-performing user segment is a more effective path to growth. Business model pivots, such as moving from paid to free, also fail to generate traction if the product does not solve a meaningful problem. Users who are indifferent to a product's value will rarely invest time in it even when the cost is removed. Successful pivots require identifying segments where the product is already working or aligning the product with existing market categories that customers understand. Founders should leverage secrets learned about the market to restart with a fresh perspective or identify successful competitors and invert a key feature to create a distinct value proposition. If an entire space shows poor retention across all players, the issue likely resides in the market itself rather than the specific implementation. This analytical approach to pivoting focuses on solving specific problems for defined user groups rather than expanding scope to mask underlying product failures.

The end of the 1 billion active user ad-supported consumer startup

Andrew Chen posits that the era of the billion-user horizontal consumer app has ended. Since the launch of TikTok eight years ago, no new horizontal giant has emerged, signaling the final stages of the mobile S-curve. Several structural hurdles now block broad-based apps. Consumer novelty has vanished and retention is difficult because new apps must compete with highly addictive, established platforms. Furthermore, the ad-supported model presents a "two miracle" challenge: achieving massive scale while simultaneously building a complex advertising marketplace. Traditional growth levers, such as viral invites and cheap mobile ads, have also been neutralized or become prohibitively expensive. The emerging alternative is the vertical app. These products target specific market segments and prioritize high monetization over massive reach. Examples include Monopoly Go, Draft Kings, and Canva. These apps often employ "whale monetization" where a small percentage of users generate the majority of revenue through direct spending or high-tier subscriptions. This model shifts the focus from total user count to Average Revenue Per Paying User (ARPPU). AI serves as a catalyst for this shift by enabling novel productivity and prosumer tools. These tools often monetize through usage-based tiers or enterprise features, creating direct revenue paths. Additionally, network effects are evolving. Instead of requiring a user's entire social circle to be present, new networks are built around specific activities or interests, such as dating or multiplayer gaming. This reduces the initial growth burden. Finally, single-user utility, enhanced by game mechanics, allows products to remain sticky without relying on social interaction.

Marc Andreessen & Andrew Chen Talk Creative Computers

Marc Andreessen on alpha nerds, moral panics, the future of AI in video games

Marc Andreessen and Andrew Chen explore the critical intersection of gaming and emerging technology, positioning the gaming industry as the primary testing ground for alpha nerd innovations such as GPUs, 3D rendering, VR/AR, and now AI. Andreessen highlights how gaming has historically served as the vanguard for high-performance hardware and complex software paradigms before they transition into broader enterprise or consumer applications. This alpha nerd effect ensures that the most demanding users stress-test new systems, creating a roadmap for mainstream adoption. A significant portion of the discussion addresses the recurring phenomenon of moral panics surrounding new technologies. Andreessen notes that these panics have historically targeted everything from electric lighting and bicycles to contemporary anxieties over AI and cryptocurrency. He critiques the prevailing tech pessimism, particularly among the Millennial generation, attributing this outlook to formative experiences during the 9/11 attacks and the 2008 financial crisis. To counter this, he advocates for a modern pro-tech political stance that engages constructively with government regulation to ensure innovation is not stifled by reactionary policies. Technologically, Andreessen defines AI as a fundamental shift toward probabilistic computing. This represents a creative paradigm distinct from the traditional deterministic computing that has defined the industry for decades. This shift enables entirely new art forms and immersive experiences, especially when integrated with gaming environments. He also emphasizes the strategic importance of open-source AI as a mechanism to prevent market consolidation by big tech incumbents, thereby democratizing access for the next generation of startups. The conversation concludes by examining how gaming mechanics and technologies can transform high-impact sectors like education and healthcare. By evolving beyond hit-driven entertainment into long-term platforms, gaming provides a blueprint for making these essential services more engaging and effective through interactive, immersive design.

Boom time startups vs Gloom time startups - by Andrew Chen

Economic cycles dictate the operational reality for startups, creating a fundamental divide between boom time and gloom time strategies. Boom times are characterized by high excitement and accessible capital, yet they present significant hidden costs. Founders face extreme competition for talent against multi-million dollar corporate packages, inflated vendor pricing, and a crowded market of similar-sounding products. This environment often forces unsustainable growth rates and limits the time available to build meaningful investor relationships, as deals are often closed in 48 hours based on hype rather than diligence. Conversely, starting during a recession or gloom time prioritizes business fundamentals and high cash efficiency. These periods filter out tourists, leaving hardcore founders to work with less distraction. Customer acquisition can actually become easier as the volume of cold outreach from competitors drops, making a targeted pitch more intriguing to potential buyers. Investors in these periods are less rushed, allowing founders to ensure a long-term cultural and strategic fit before adding them to the cap table. Historical context proves that building during a downturn often precedes massive success. The period between 2002 and 2007 allowed founders to experiment with the new physics of the internet, leading to the rise of platforms like YouTube, Yelp, and LinkedIn. These companies were built with the discipline required in a post-bubble environment, positioning them perfectly for the subsequent mobile and social explosion. Choosing a strategy depends on founder personality and the nature of the problem. Some ideas require the high-burn, high-growth momentum of a boom, while others benefit from a decade-long grind toward becoming a foundational SaaS entity. Current economic turbulence should be viewed as an opportunity to build with strong foundations, as the most impactful startups are often those that survive the lean years to ride the next wave.

Why high growth, high churn products never seem to work

Andrew Chen examines the phenomenon of meme apps, which are products characterized by explosive viral growth followed by rapid abandonment. These apps often leverage AI to create single-use, highly visual experiences that thrive on social media platforms like TikTok and X. While these products frequently reach the top of app store charts, they often fail because they lack sustainable retention. The core issue is mathematical: while acquisition typically follows a linear or S-curve path, churn operates as a percentage of the total active user base. This creates a crossover point where the volume of departing users eventually exceeds the number of new signups, leading to a permanent decline in active users. The text warns against the Illusion of Success, where temporary viral spikes mask poor underlying product-market fit. Spiky acquisition is difficult to repeat and often attracts low-intent users who do not convert into long-term value. For networked products, such as marketplaces or social platforms, these disconnected spikes fail to build the foundational density required for strong network effects. To achieve the T2D3 growth trajectory (tripling and doubling revenue over several years), a product must have high retention. Without it, a company enters a growth treadmill where it must exponentially increase its acquisition efforts just to replace the users it loses annually. As the novelty window for AI technology closes, founders must transition from riding viral waves to building utility. This involves identifying sticky user segments, integrating into existing professional workflows, and developing collaboration features. The evolution of the AI market will likely mirror previous tech cycles, moving from disposable novelty to platform-based utility where technical capabilities are secondary to integrated user value.

The case against morning yoga, daily routines, and endless meetings

Professional output follows a power law distribution where a minority of high-impact tasks, defined as 10x work, drive the vast majority of career value. Most individuals remain trapped in 1x work, which includes repetitive routines, recurring meetings, and administrative checklists. While these activities offer a sense of machine-like efficiency and comfort, they lack the risk and information density required for exponential growth. The root cause of productivity stagnation is often an over-reliance on these defensive core loops and reactive behaviors, such as managing an overflowing inbox, rather than pursuing high-variance opportunities. 10x work typically occurs at the frontier of knowledge, such as AI, web3, or deep tech, where status hierarchies are not yet established and the right way of doing things remains unwritten. To maximize these moments, one must prioritize agency and serendipity over routine. This involves a fundamental shift from reactive replies to proactive outbound actions. Examples include publishing original thoughts publicly, hosting high-signal events, and building compounding assets like code or content that scale non-linearly. These activities increase the surface area for luck and allow work to compound even while the creator is inactive. Strategic advantage is further enhanced through the concept of skill stacking. Rather than attempting to be the absolute best in a single, highly competitive field, individuals should aim to be in the top 25% across multiple complementary domains. This combinatoric approach creates a rare and valuable intersection that is difficult for competitors to replicate. Ultimately, achieving 10x results requires a dual-track strategy: playing defense by automating or delegating low-value 1x tasks, and playing offense by aggressively investing in comparative advantages and high-upside projects that force rapid learning through tests of skill.

Why Hollywood and gaming struggle with AI - by Andrew Chen

The entertainment and video game industries are facing significant friction in adopting generative AI due to structural, legal, and cultural barriers. While executives acknowledge AI's potential to reinvent their fields, the innovator's dilemma prevents proactive adoption. Large companies with established franchises are reluctant to risk existing business models or workflows for unproven technology. When incumbents do engage with AI, they typically bolt it on as a backend cost-saving feature rather than reinventing products from an AI-first perspective. This mirrors the slow adoption of mobile and free-to-play models in gaming. For instance, major titles like Call of Duty and Diablo did not release mobile versions until a decade after the platform's emergence. Legal hurdles regarding training data and IP ownership represent a major bottleneck. Corporate legal departments often restrict product teams because AI models may inadvertently produce copyrighted content. In contrast, startups are more likely to prioritize product-market fit and consumer demand, dealing with legal complexities as they scale. This approach allows startups to validate whether consumers actually want AI-generated content before building defensive legal frameworks. Furthermore, the creative workforce in these industries, including artists and writers, often views AI as a threat to their livelihood. The high ratio of creative staff to engineers in these firms creates a powerful internal lobby against automation, as seen in the recent SAG-AFTRA strikes. The difficulty of recruiting AI talent further compounds these issues. Top-tier AI researchers command multi-million dollar salaries, making it nearly impossible for established entertainment firms to compete with tech giants or high-upside startups. This talent adverse selection forces incumbents to rely on external vendors, further slowing innovation. However, this friction creates a massive opening for startups to build AI-native products. These new ventures will likely move beyond existing formats to create ephemeral, niche, or meme-like experiences that target specific world events or small audiences. The ultimate evolution is the $1,000 blockbuster movie, where AI provides 100,000x leverage, allowing individuals to produce high-quality films or games with minimal capital. This democratization will likely lead to an explosion of content and new genres, similar to how user-generated content transformed the Web 2.0 era.

Why data-driven product decisions are hard (sometimes impossible)

Data-driven decision making faces a central paradox: while intended to be objective, the interpretation of metrics is frequently clouded by subjective excuses like seasonality, time lags, or conflicting A/B test results. This "fog of war" makes truth-seeking difficult during product reviews and board meetings. Several root causes undermine the purely data-driven approach. First, data typically reflects yesterday's market and early adopters rather than the broader future market. This often leads teams to chase a local maximum by building features for a niche audience that do not scale to a mass market. Second, the high cost of data analysis often results in analysis paralysis. When the cost of acquiring and analyzing data is too high, the speed of expert intuition becomes a more valuable asset. Third, A/B tests are structurally biased toward measuring high-volume, short-term actions like signup rates or clickthrough rates. They often fail to capture 90-day retention for high-value minority segments, leading teams to ignore the most important users. Furthermore, noise from external factors like ad campaigns or competitor launches makes clean comparisons difficult. The common confusion between correlation and causation also leads to wasted effort on features that do not actually drive engagement. In established tech cultures like Google or Meta, a data-driven approach supports incremental optimization. However, startups competing against these giants cannot win by playing the same game. Startups must prioritize intuition and rapid, large-scale moves over slow, incremental data-backed decisions. Transitioning from being data-driven to data-informed, or even data-ignorant in zero-to-one phases, allows for expert qualitative judgment to guide strategy before optimization becomes appropriate.

Time sinks and money sinks - by Andrew Chen - @andrewchen

Internet products generally fall into two categories: time sinks and money sinks. Time sinks prioritize high engagement and stickiness, often living on the user's home screen. These apps, such as social networks and news sites, are typically free or low cost and exhibit poor monetization on a dollar per hour basis. Conversely, money sinks are low engagement and episodic. Users visit these platforms to complete a specific task, such as buying a car or securing a mortgage, and are willing to spend significant capital. Because money sinks struggle with organic retention, they frequently purchase traffic from time sinks. Business Pain / Root Cause: The primary challenge is the metrics mismatch. Founders often default to high frequency engagement metrics like DAU or D1 retention for transactional businesses where CAC/LTV and conversion efficiency are the true drivers of value. This memetic thinking leads to generic product strategies that fail to optimize for the actual business model. The New York Times serves as a leading indicator for the All Access bundle trend. By integrating news, cooking, and gaming, they are pivoting toward a model where gaming provides the primary retention hook while news provides cultural relevance. This mirrors moves by Netflix and Amazon, suggesting a future where media companies converge on a similar package of video, music, and interactive entertainment. Product design is a downstream variable of the business model. The shift from arcade quarter eaters to free to play models demonstrates that difficulty levels, content depth, and social features are dictated by how the company captures value. As subscription models dominate, the incentive shifts from driving clicks via sensationalism to maintaining a steady stream of revenue through adjacent, high engagement categories like gaming.

How the coupon was invented, and how tech propels marketing

Technology propels marketing through a recurring cycle where every widely adopted innovation eventually becomes a vehicle for promotion. This relationship is best illustrated by the invention of the coupon in 1887 by Asa Candler for Coca-Cola. While often viewed as a simple consumer discount, the coupon was actually a sophisticated solution to a B2B2C cold start problem. By providing free syrup to retailers and simultaneously distributing coupons to local consumers, Coca-Cola stimulated both supply and demand. This tactic was only possible due to the convergence of the steam-powered printing press, chromolithography, and the US Postal Service's implementation of free home delivery. These technologies created the necessary 'surface area' for mass-market physical marketing. In the modern era, promo codes have become the digital successor to coupons, yet they face the same fundamental challenges: demand cannibalization, fraud, and attribution. Marketing teams, such as those at Uber, have spent billions managing these incentives by measuring 'Cost Per Incremental Trip' to distinguish between new growth and purchases that would have occurred regardless of the discount. This highlights a persistent tension in marketing: the ease of driving top-line revenue versus the difficulty of maintaining high-quality, profitable growth. Looking forward, AI companions and interactive agents represent the next major surface area for marketing. As users shift toward conversational interfaces, marketing will move from static displays to native, stateful interactions. This transition will likely give rise to 'AI SEO,' where companies optimize their content specifically to be incorporated into LLM training data to influence agent recommendations. Furthermore, as these agents maintain long-term context and user preferences, they will facilitate real-time loyalty negotiations and cross-selling. Despite the current prevalence of subscription models for AI, historical patterns suggest a shift toward free, ad-supported models is inevitable as consumers prioritize accessibility over privacy.

How AI will reinvent Marketing - by Andrew Chen

AI's impact on marketing will move beyond incremental efficiency to structural reinvention. This shift is characterized by the transition from horseless carriages to entirely new urban landscapes. The root cause of current marketing limitations is the high cost of human labor, which forces brands into broadcast models. AI solves this by transmuting capital into infinite labor via compute. This enables infinite content production where costs approach zero, allowing for mass personalization and real-time updates to brand campaigns based on sentiment. Internationalization becomes instantaneous, moving beyond simple translation to deep cultural localization of imagery and product UX. The user experience shifts toward a concierge model where onboarding and support provide five-star service through empathetic AI agents. Furthermore, the depth of promotional content will explode; instead of video trailers, brands may deploy entire video games or TV seasons as interchangeable digital assets. Marketing channels will face significant upheaval as LLMs become the primary interface, potentially making traditional SEO and SEM obsolete. These will be replaced by AI companions and sponsored recommendations within voice and chat conversations. This accelerates the OODA loop (Observe, Orient, Decide, Act) from months to minutes, allowing brands to adjust strategies instantaneously. A critical evolution is the convergence of marketing and sales. Historically, marketing was a broadcast tool because 1:1 sales were too expensive. AI removes this constraint, enabling millions of personalized 1:1 conversations that function as a virtual salesforce. Finally, as AI-generated content achieves perfect beauty and polish, a counter-trend will emerge. A premium will be placed on authenticity and Proof of Human, where intentional imperfections signal real human connection in a saturated digital environment.

10 years after "Growth Hacking" - by Andrew Chen

The discipline of growth hacking has transitioned from an era of abundance during the early mobile S-curve to a period of scarcity and saturation. A decade ago, the novelty of the App Store and high deliverability of social invites allowed for rapid organic growth. Today, the mobile ecosystem is mature, and developers compete against highly addictive, established platforms, leading to a stagnant base rate of new growth. This shift highlights the inherent limitations of growth hacking tactics, particularly for early-stage startups. A/B testing and incremental optimizations are insufficient for achieving product-market fit, as they cannot compensate for poor retention or fundamental flaws in product strategy. Strategic outcomes are largely dictated by the Nature vs Nurture framework, where the product category and chosen customer segment determine baseline retention and frequency metrics. Tactical nurturing through growth hacking can only move these metrics marginally. Furthermore, aggressive growth projects often result in UX cruft, which degrades long-term engagement in favor of short-term gains. Despite these challenges, the growth mindset has been institutionalized, with product managers and marketers now expected to be fluent in distribution metrics like CAC, LTV, and D30. The next frontier of growth is driven by the AI S-curve and video-native products. Generative AI reduces the cost of building and allows for it actually works novelty that triggers organic sharing. Most significantly, AI enables a shift from 1:many marketing broadcasts to mass 1:1 personalized sales agents. This transition allows companies to deploy virtual salesforces that provide white-glove, concierge experiences at a fraction of the cost of human labor. As marketing and sales converge through AI-powered personalization, the focus shifts from optimizing flows to leveraging AI as a core distribution lever.

I've hit the 6 year mark at a16z! - by Andrew Chen

Andrew Chen reflects on six years at Andreessen Horowitz, detailing his transition from a broad consumer focus to a specialized role at the intersection of gaming and technology. The firm has expanded from 100 to 600 employees and recently raised $7.2 billion, including a dedicated $600 million Games Fund. This fund targets the $300 billion gaming industry, focusing on AI, infrastructure, web3 games, and 3D tooling. Chen highlights the SPEEDRUN accelerator, which provides $750,000 in funding for early-stage startups. A significant portion of the update focuses on the AI supercycle and its potential to transform interactive entertainment. Chen argues that while incumbents in Hollywood and gaming face the Innovator's Dilemma, struggling with legal IP concerns, creative workforce pushback, and the difficulty of hiring AI talent, startups have a unique opportunity to build AI-native experiences. These new products may be more ephemeral, niche, or meme-like, deviating from traditional high-production-value models. The move to Los Angeles and the opening of a Santa Monica office signal the firm's commitment to the local ecosystem, including participation in LA Tech Week. Chen concludes that AI will likely follow the path of previous gaming-originated technologies like GPUs and streaming, eventually dominating the broader consumer tech landscape.

How novelty effects and Dopamine Culture rule the tech industry

The transition from traditional slow culture to digital Dopamine Culture has fundamentally altered product engagement and growth dynamics. Traditional culture relied on high-intent, offline experiences with long production cycles and restricted supply. In contrast, Dopamine Culture prioritizes instant accessibility, endless variety, and ephemeral digital interactions. This shift is driven by universal smartphone access and the removal of central gatekeepers, allowing niche interests to flourish through handheld, mobile supercomputers. For tech builders, this environment creates a landscape where moments are measured in seconds rather than days. The primary business pain resulting from this culture is the ubiquity of poor retention. Standard benchmarks for successful consumer apps, such as Day 1/7/28 retention of 60/30/15 and a DAU/Registered ratio of 25%, reveal that losing 90% of daily active users within a month is considered a high-tier performance. The root cause is a lack of friction; users can instantly abandon a product for a more stimulating alternative if value is not delivered in the first session. Consequently, the time to value must be near-zero to prevent immediate churn. The Novelty Effect further complicates growth strategies, particularly within the AI sector. As new technology S-curves emerge, they trigger intense dopamine responses, leading to high organic word-of-mouth and signup conversion. However, this often results in spiky growth curves and one-and-done usage. Product management culture has adapted to these trends through quarterly OKR cycles, which frequently lead to the Next Feature Fallacy. This fallacy assumes that incremental features or algorithmic tweaks will solve fundamental retention issues, when they often only provide temporary engagement spikes. Success in this environment requires one of two strategic paths. The first is to lean into Dopamine Culture by creating media-centric, fast-value loops that naturally go viral. The second is counterpositioning, where high-dopamine, short-form content acts as a commercial for deep-engagement products like Substack or long-form podcasts. While riding the novelty curve provides an initial advantage, long-term viability depends on transitioning from a novel experience to a demonstrably useful one before the S-curve matures.

(6) The mobile S-curve ends, and the AI S-curve begins

The technology landscape is currently defined by the intersection of a maturing mobile S-curve and a nascent AI S-curve. Mobile, having launched fifteen years ago, is now a stable duopoly where the top one hundred apps rarely change. Success in this late-stage environment requires radical differentiation or superior design to overcome high user expectations. Conversely, the generative AI sector is in a state of productive chaos. Startups in this early phase benefit from the "It Works" feature, where the sheer difficulty of building functional technology creates immediate demand and viral waitlists. This growth is often accelerated by visual outputs that are easily shared on social media, creating a feedback loop of high-level growth numbers. However, early S-curve growth is frequently deceptive. Chen argues that humans are "dopamine fiends" who respond to the novelty of AI-generated content with high engagement and sharing. As users habituate to these outputs, the novelty effect unwinds, leading to high churn. Retention becomes the primary challenge as the market moves up the S-curve. Investors favor this early stage because the "rising tide" of the market allows for exponential growth even with unrefined products. This leads to opportunistic pivots and intense competition, which Chen views as a sign of a healthy, high-value market. If you are in a market with no competition, it may not be a great market. In contrast, late S-curve markets like mobile require a different skill set. Growth channels are saturated by established incumbents in gaming, travel, and ecommerce, making customer acquisition expensive and difficult. Startups in these "red oceans" must adopt an "anti-incumbent" thesis, such as BeReal's challenge to Instagram, or leverage a design-led "last-mover advantage." This approach involves entering a market after the initial technical hurdles are solved to perfect the user experience, a strategy historically mastered by Apple with the GUI and the smartphone. Founders must recognize which phase of the S-curve they are building on to properly calibrate their product development, marketing efficiency, and expectations for growth.

Building the initial team for seed stage startups at andrewchen

Seed stage startups transitioning from founder-led operations to a core team of 4 to 6 employees face significant structural risks. The primary challenge is the specialist trap, where hiring for narrow roles limits the agility required for early-stage discovery. Instead, founders should prioritize T-shaped individuals who possess broad contextual knowledge and deep expertise in one critical area. This versatility allows for seamless cross-functional communication and ensures the team can adapt when the startup inevitably changes direction. The root cause of poor hiring at this stage is often a lack of candidate flow, which leads founders to lower their standards to fill roles quickly. Establishing a repeatable sourcing process in specific communities is the only sustainable solution to maintain a high talent bar and ensure that the first batch of employees is of the highest caliber. Execution focus is significantly more valuable than seniority or strategic background during the seed phase. Hiring doers, specifically those with recent experience as team leads or directors, is preferable to hiring senior executives, consultants, or philosophers who may prioritize intellectual indulgence over concrete product development. These profiles often struggle with the immediate, hands-on requirements of a startup environment. Furthermore, traditional interview methods are often insufficient for predicting actual job performance. A three-day working interview provides a high-fidelity signal of how a candidate actually performs in a collaborative setting. Finally, raw intelligence must be balanced with intrinsic motivation and cultural alignment. The distinction between missionaries, those aligned with the product mission, and mercenaries, those focused solely on metrics or compensation, is critical. Missionaries provide the cultural bedrock and resilience needed for the seed stage, whereas mercenaries may churn during periods of volatility. Metrics-oriented individuals should ideally be trained to love the product, rather than attempting to force a data-first hire to care about the mission after the fact.

How To (Actually) Calculate CAC at andrewchen

Accurate Customer Acquisition Cost (CAC) calculation requires a fundamental distinction between CAC and Cost Per Acquisition (CPA). CPA measures the cost of leading indicators such as registrations, leads, or trials, whereas CAC specifically tracks the cost to acquire a paying customer. Using these terms interchangeably leads to flawed growth projections and incorrect company valuations. The standard formula of total marketing and sales expenses divided by new customers is often misleading because it fails to account for three critical variables: timing, expense depth, and customer definition. Timing is the most frequent source of error. In B2B SaaS or freemium models, a significant delay exists between the initial marketing spend and the final conversion. For instance, if a lead takes 60 days to close, calculating CAC based on the current month's spend will result in an inaccurate figure. This lag can cause teams to prematurely shut down effective channels that appear expensive in the short term but yield high-value customers later. To fix this, expenses must be time-shifted to align with the average sales cycle length. Calculating a "Fully Loaded" CAC is essential for operational precision. This includes not only direct ad spend but also salaries of marketing and sales staff, overhead costs like rent and equipment, and the stack of software tools used to manage the funnel. In Product-Led Growth (PLG) environments, the definition of acquisition costs may even extend to engineering and support if the free product tier is the primary driver for new user acquisition. Finally, the calculation must strictly separate new customers from returning or reactivated users. Including returning users in the denominator without accounting for retention spend in the numerator artificially deflates the CAC, providing a false sense of efficiency.

A Practitioner’s Guide to Net Promoter Score at andrewchen

Net Promoter Score (NPS) serves as a primary KPI for measuring customer loyalty and predicting word-of-mouth virality. Sachin Rekhi, drawing from experience at LinkedIn, emphasizes its utility in understanding customer delight beyond standard acquisition and monetization metrics. Devised by Fred Reichheld, the metric categorizes users into Promoters (9-10), Passives (7-8), and Detractors (0-6) based on their likelihood to recommend a product. The final score is derived by subtracting the percentage of detractors from the percentage of promoters. While the numerical score provides a benchmark, the qualitative "Why" question is the essential driver for product improvement. Effective NPS implementation requires rigorous sampling to avoid bias. Responses often correlate with user engagement and tenure. Therefore, the sample must mirror the actual user base to ensure accuracy. Collection should occur across desktop and mobile interfaces via email or in-product prompts. A quarterly survey cadence typically aligns best with product planning cycles, allowing teams to integrate findings into upcoming roadmaps. The primary business pain addressed by NPS is the lack of visibility into customer sentiment beyond basic engagement metrics. Actionable insights come from verbatim analysis. Categorizing open-ended comments from both promoters and detractors reveals primary benefits and root causes of dissatisfaction. Analyzing promoter behavior, including frequency of searches or profile views, helps identify the magic moment where users derive maximum value. This allows teams to optimize the product to lead more users toward these high-value actions. Despite its utility, NPS has limitations. It is a lagging indicator and unsuitable for day-to-day operational tracking or A/B testing. It also carries a margin of error dependent on sample size. NPS should be viewed as a tool to evaluate execution against a product strategy rather than a replacement for the strategy itself. Consistency in methodology, including question ordering and competitor benchmarking, is critical for maintaining comparable data over time.

The red flags and magic numbers that investors look for in your startup’s metrics – 80 slide deck included! at andrewchen

Evaluating startup viability requires moving beyond vanity metrics to understand the underlying physics of growth. The Growth Accounting Framework defines Net Monthly Active Users (MAU) as the sum of New and Reactivated users minus Inactive users. Growth plateaus occur naturally when the acquisition of new users, which typically follows a linear or S-curve, is equaled by the lagging curve of inactive users. To predict future performance, investors must analyze two primary types of loops: Acquisition and Engagement. Acquisition loops are proprietary, repeatable, and scalable systems such as UGC SEO (e.g., Yelp, Glassdoor), paid marketing, and viral loops. Unlike linear channels like PR or conferences, loops amplify every new user through a cycle of reinvestment or invitation. The efficiency of these loops is measured by the viral factor; for example, a ratio of 0.6 means 1,000 users eventually generate 1,500 total signups. Red flags in acquisition include high platform dependency, such as building on a shrinking third-party ecosystem, or 'juiced' metrics from unsustainable ad spend spikes before fundraising. Engagement loops focus on retention through social feedback or 'planting seeds' where early user actions trigger future re-engagement (e.g., Zillow property alerts). High-quality engagement is evidenced by cohort retention curves that flatten above 20%. If curves do not flatten, the product lacks a sticky core. Investors should also look for a 'ladder of engagement,' where users move from low-frequency use cases to high-frequency habits. Identifying 'upside' involves decomposing these loops into granular steps to find friction points, such as high bounce rates on landing pages or failed login attempts. A robust forecast combines these acquisition and engagement scenarios into a bottom-up model rather than relying on historical spreadsheet trends.

Required reading for marketplace startups: The 20 best essays at andrewchen

Marketplace success relies on navigating the transition from initial liquidity to sustainable network effects. The chicken and egg problem remains the primary barrier to entry. Effective strategies to overcome this include liquidity hacking through geographic or vertical narrowing, providing single-player value via tools or community, and leveraging existing supply aggregators. Modern marketplace evolution favors managed or full-stack models. These platforms move beyond simple lead generation to control the end-to-end user experience, including supply-side software, pricing guidance, and trust mechanisms. This shift addresses the limitations of traditional search-based models by improving retention and transaction frequency. The Market Network concept represents a significant structural trend. These platforms combine the transactional nature of marketplaces with the workflow efficiency of SaaS tools. Unlike standard marketplaces where participants are often interchangeable, market networks focus on complex services where individual profiles and long-term relationships are critical. Quantitative evaluation of marketplace health requires a rigorous focus on specific KPIs. Beyond Gross Merchandise Volume (GMV), operators must track liquidity metrics, cohort retention, and unit economics such as LTV/CAC ratios. Successful marketplaces typically exhibit high fragmentation on both supply and demand sides, high transaction frequency, and a significant value proposition compared to the status quo. Strategic failure often stems from a lack of perfect competition or failing to account for the hyperlocal nature of network effects in specific industries.

Growth Interview Questions from Atlassian, SurveyMonkey, Gusto and Hubspot (Guest Post) at andrewchen

Hiring for growth roles requires a balance of creative problem-solving and rigorous, metrics-based iteration. This analysis synthesizes interview strategies from leaders at Atlassian, SurveyMonkey, Gusto, and HubSpot. Key tactics include using "weird" questions to observe reasoning under uncertainty, such as Elena Verna’s Golden Gate Bridge estimation task. Nick Soman utilizes first-principles thinking exercises, like asking how to growth hack a city, to see how candidates operate without established playbooks. Shaun Clowes focuses on early adopter behavior to gauge industry intuition. Andrew Chen advocates for a "Bullshit Test" where candidates must whiteboard the growth loops of a known product in real-time to demonstrate technical depth beyond industry jargon. A significant shift in strategy involves hiring for "superpowers" and long-term trajectory rather than simply filling immediate "urgent holes" or fixing leaks. The framework also highlights the importance of assessing resilience through personal setbacks. For candidates, the guide suggests probing into the 6-month roadmap, resource autonomy, specifically dedicated engineering and design support, and the history of homepage optimization to determine if the "low-hanging fruit" has already been exhausted.

Why “Uber for X” startups failed: The supply side is king at andrewchen

The widespread failure of Uber for X startups is primarily attributed to a failure to master supply side economics. While the rideshare model is often emulated, its success relies on a specific balance of high acquisition costs and high labor utilization. Uber spends significantly to acquire drivers, often upwards of 300 dollars per person, but recovers this investment because drivers can work 20 to 50 hours per week with consistent demand. The driver app remains a constant necessity for the worker to find new customers throughout the day. Most Uber for X companies in sectors like valet parking, car washing, or massage therapy fail because demand is infrequent and concentrated in narrow time windows. This creates a structural problem where the supply side remains idle during off-peak hours, yet the cost of labor and acquisition remains as high as in the rideshare industry. Consequently, these businesses struggle to achieve positive unit economics and remain unprofitable even at scale. Standalone food delivery businesses face similar challenges. They must compete for the same pool of drivers but often lack the diversified demand required to pay them competitively. Uber maintains an advantage by augmenting driver earnings with both rideshare and delivery requests, effectively subsidizing the delivery arm through a more utilized supply side. To build a successful marketplace, founders must move beyond simple emulation and return to first principles. This involves identifying untapped labor pools, such as individuals who prefer working from home, those in rural areas, or people without vehicles. The most defensible marketplaces are those that solve fragmentation through transparency and aggregation while ensuring the supply side can remain active and sticky for a full work week. Evaluating a marketplace opportunity requires looking through the lens of the worker. If the platform cannot provide consistent, full-time earning potential, it will likely struggle with supply constraints and eventual failure.

Uber’s virtuous cycle. Geographic density, hyperlocal marketplaces, and why drivers are key at andrewchen

Uber operates as a vast collection of hundreds of hyperlocal two-sided marketplaces across nearly 70 countries rather than a single global network. Each local market relies on a virtuous cycle driven by geographic density, a concept popularized by David Sacks and expanded upon by Bill Gurley. This cycle is powered by three primary factors: pick-up times, coverage density, and utilization. As demand and supply grow within a specific city, pick-up times decrease, making the service more reliable and opening up more use cases. Simultaneously, coverage density expands from city centers to outskirts, increasing supplier liquidity across a wider geographic range. Higher demand leads to better driver utilization, solving the complex mathematical problem of minimizing idle time between trips. This increased efficiency allows Uber to lower fares, which further stimulates demand and reinforces the cycle. The supply side of the marketplace is the most critical operational pillar. Uber manages over one million active drivers globally, but the labor market is characterized by significant volatility. Data indicates that while many people use the platform for supplemental income to cover unexpected expenses, retention is a major challenge. Only about 40 percent of drivers remain active one year after their first trip. This high churn necessitates a constant, aggressive focus on driver acquisition to maintain the network effect. Marketplace balancing mechanisms like surge pricing and fare cuts are essential tools for managing this volatility. Surge pricing acts as a hyperlocal incentive, using real-time data to direct drivers to specific high-demand hexagons within a city. Conversely, fare cuts are strategically used to boost demand during slow periods. When paired with temporary earnings guarantees, these cuts can actually increase driver earnings-per-hour by reducing downtime and increasing trips-per-hour. Ultimately, Uber's success depends on its ability to maintain equilibrium between riders and drivers at a local level through these algorithmic and economic interventions.

The Next Feature Fallacy: The fallacy that the next new feature will suddenly make people use your product at andrewchen

The Next Feature Fallacy describes the mistaken belief that building additional features will solve fundamental engagement and retention issues. This mindset often leads to a cycle of launching and failing because it ignores the underlying metrics of the user lifecycle. The core of this issue is represented by the tragic curve, a retention diagram showing a precipitous drop-off from initial signup to Day 30. In a typical scenario, 1000 homepage visitors might result in only 20 daily active users after one month. This steep decline occurs because most users drop off before ever experiencing the product's core value. Most features fail to bend this curve for two reasons. First, they often target already engaged or retained users rather than the much larger pool of non-users and new users. Second, the features often lack sufficient impact even when users do engage, frequently because they are hidden outside the primary onboarding flow. A Day 7 feature, for instance, will only be seen by the small fraction of users who survive the first week, limiting its potential to influence overall growth. To combat this, product teams must address the engagement wall. This is the point where a product asks for a deep investment, such as creating a project or uploading files. Features behind this wall are only experienced by a small percentage of users. Conversely, features in front of the wall provide value with minimal investment. Effective growth strategy requires a strong opinion on user activation. For example, Twitter shifted from a blank feed to a forced-follow onboarding process to ensure users immediately experienced the product as a reader. Maximizing reach by focusing on the landing page, onboarding sequence, and initial out-of-box experience is the most reliable way to drive meaningful improvements in retention.

Why consumer product metrics are all terrible at andrewchen

Consumer product metrics for signup, retention, and engagement are frequently lower than industry newcomers anticipate. Most products face a reality where 90% of visitors decline to sign up, and 90% of those who do eventually disengage. This analysis breaks down the structural challenges across three primary areas: signup rates, retention frequency, and social graph density. Signup rates vary significantly by traffic source. While minimal homepages can convert at 20% to 30%, SEO-driven landing pages often convert at less than 1% because visitors prioritize specific content over the platform itself. Retention curves typically show a steep drop-off within the first 14 days before stabilizing at single-digit percentages. Achieving a 10% daily return rate is considered a success, though top-tier unicorns like WhatsApp reach 70%. Engagement is largely a function of the product category rather than superficial optimizations. If a product lacks inherent frequency, cosmetic changes like push notifications or email reminders rarely improve long-term DAU/MAU. Social products also struggle with empty feeds. Even Instagram saw 65% of its users follow effectively no one during its first year. Solving this density problem requires leveraging existing social graphs or forcing follows during onboarding to ensure a functional content feed. While these metrics appear mediocre, successful businesses often scale by monetizing the small percentage of users who remain active and engaged.

New data shows losing 80% of mobile users is normal, and why the best apps do better at andrewchen

Mobile app retention data from over 125 million devices reveals that the average application loses 77% of its daily active users (DAUs) within the first three days of installation. By the 30-day mark, 90% of users are lost, and after 90 days, the loss exceeds 95%. This data, provided by mobile intelligence startup Quettra, highlights that the vast majority of the 1.5 million apps in the Google Play store fail to sustain meaningful traffic. The critical window for user retention occurs within the first 3 to 7 days. Users typically decide which apps to stop using almost immediately, meaning the most significant leverage for improving retention lies in the initial onboarding flow and the product's value proposition rather than long-term re-engagement efforts. Comparison between top-performing apps and average apps shows that the highest-ranked applications maintain significantly higher D1 retention rates. While the rate of decay from day 1 to day 30 is relatively consistent across all apps, the top 10 apps start with a much higher baseline, ending with approximately 60% retention at day 30 compared to less than 10% for the average app. This suggests that users find top apps immediately useful and integrate them into their routines during the first week. Bending the retention curve requires focusing on activation milestones rather than notification-based re-engagement. Effective activation strategies include getting users to pick a theme in a blogging tool, connecting friends in a social service, or installing a Javascript tag in a SaaS analytics product. These actions create user investment and 'hook' the audience during the vital first visit.

How startups die from their addiction to paid marketing at andrewchen

Startups frequently collapse due to an unsustainable reliance on paid marketing channels while misinterpreting Customer Acquisition Costs (CAC). This Paid Marketing Local Max occurs when initial success with early adopters leads to aggressive spending and venture capital funding, only for growth to hit a ceiling as unit economics deteriorate. A critical error is focusing on Blended CAC, which averages organic and paid users. Because early organic users are often the most loyal, Blended CAC can be 2 to 5 times lower than actual channel-specific CAC. As paid spend scales, the Blended metric inevitably converges with the higher cost of the dominant paid channel. The root cause of this addiction is often a psychological reliance on immediate, trackable growth that masks underlying product weaknesses. Scale effects in paid marketing generally work against the advertiser. Creative fatigue, message staleness, and the reversion to the mean reduce effectiveness over time. Furthermore, as companies exhaust their core demographic, they must target less responsive audiences, increasing costs. Competitive dynamics also play a role. Rivals can easily replicate ad copy and landing pages, whereas viral mechanisms like Dropbox's folder sharing or Slack's channel creation are deeply integrated into the product and harder to copy. Paid marketing is strategically justified in specific scenarios, such as bootstrapping network effects or when a company possesses a deep algorithmic edge in adtech integration. However, for most, it should be capped at 30% to 40% of top-of-funnel activity. Sustainable growth requires building secondary and tertiary channels, including referral programs, content strategies, and viral product features. The Dropbox case study illustrates this shift. After finding paid search unprofitable in a mature market, the company pivoted to viral loops that aligned with user behavior and drove exponential growth. Long-term success depends on fixing underlying issues like churn and frequency rather than extending the LTV window to justify rising acquisition costs.

Benefit-Driven Metrics: Measure the lives you save, not the life preservers you sell at andrewchen

Value creation is the primary driver of revenue and traffic, yet most standard analytics focus on inward-looking metrics like account registrations, pageviews, and monthly unique visitors. These metrics prioritize value extraction for the business rather than value creation for the customer. Optimizing for internal KPIs often leads to short-term gains but long-term product degradation. The core framework introduces Benefit-Driven Metrics, which require measuring the specific value a customer receives from a product. Using the analogy of a life preserver company, a business focusing on sales might optimize for cheaper manufacturing or aggressive sales tactics. Conversely, a business focusing on lives saved would optimize for distribution and effectiveness. While the actions might look similar, the underlying spirit and long-term outcomes differ significantly. To implement this, organizations must identify what their customers are measuring. For dating sites, this means tracking successful matches rather than lifetime value (LTV). For marketplaces, it involves measuring seller earnings and buyer satisfaction rather than listing fees. Social networks should prioritize meaningful interactions, such as posts that receive replies, instead of simple user registrations. Online publishers should focus on the conversions or revenue generated for advertisers rather than their own CPMs. Ad-supported startups face a unique challenge because their true customer is often the advertiser, not the end user. Over-focusing on user experience at the expense of advertiser ROI can be detrimental. Establishing these metrics early is vital because a startup's organizational DNA, technology, and culture become hard-coded to optimize for its initial KPIs. Transitioning from a high-churn, high-volume model to a value-driven model is difficult once the organization has matured around the wrong incentives.

DAU/MAU is an important metric to measure engagement, but here’s where it fails at andrewchen

DAU/MAU, the ratio of daily active users to monthly active users, serves as a primary benchmark for user engagement. Popularized by Facebook's consistent 50% performance, this metric primarily favors high-frequency social and messaging products. It often fails to capture the value of episodic or low-frequency products that are nonetheless highly successful. Episodic usage patterns are common in multi-billion dollar companies like Uber, Airbnb, and enterprise SaaS platforms such as Salesforce or Workday. These businesses thrive with lower DAU/MAU ratios because they generate significant revenue per transaction or accumulate high-value data. Attempting to force a higher DAU/MAU through push notifications or emails often yields counterproductive results. These tactics typically inflate MAU faster than DAU, effectively lowering the ratio. Product categories have a natural cadence that is difficult to alter through marketing alone. Instead of focusing on the aggregate ratio, companies should measure their hardcore user base or demonstrate how usage frequency correlates with network effects. The chosen metric must align with the product's natural utility and monetization model. Ad-supported businesses require high frequency, while transaction-based models can succeed on episodic value.

My Quora answer to: How do you find insights like Facebook’s “7 friends in 10 days” to grow your product faster? at andrewchen

Identifying growth levers like Facebook's "7 friends in 10 days" requires a structured approach to data analysis and organizational alignment. The process begins with defining a success metric tailored to the specific business model. For ad-supported platforms, this often involves frequency of engagement or active days within a 28-day window. For transactional models, revenue or purchase frequency serves as the primary evaluator. Once the success metric is established, practitioners must conduct cohort analysis. This involves tracking users who joined within a specific timeframe and mapping their behaviors against the success metric. Key variables to track include content creation, app downloads, social interactions, and comments received. The analysis phase involves running correlations to identify which behaviors most strongly associate with successful users. While formal regression analysis can provide statistical rigor, startups frequently encounter challenges such as insufficient data or an excess of variables. In these environments, a statistically perfect model is often less valuable than a simple, directional metric that the growth team can easily understand and act upon. The core objective is to formulate a hypothesis about what drives user success while remaining cautious of the distinction between correlation and causation. Verification is a critical step in the framework. Teams must utilize A/B testing to prioritize the identified input variable and observe if it results in a measurable increase in the success metric. If the experiment fails to show a significant difference, the model must be refined or abandoned. The final phase is branding the model. A growth metric only becomes effective when it is simplified and communicated repeatedly across the organization. This simplification ensures that the product roadmap remains focused on moving the specific lever that has been proven to drive growth.

The Power User Curve: The best way to understand your most engaged users at andrewchen

The Power User Curve, also known as the activity histogram or L30, provides a granular view of user engagement by plotting the number of days users are active within a specific timeframe. Unlike the DAU/MAU ratio, which collapses engagement into a single percentage, this histogram reveals the variance between casual and power users. A smile curve, where the right side shows a spike in daily active users, typically characterizes successful social platforms and ad-supported models. Conversely, a left-weighted curve indicates infrequent usage, which is common for professional networks or investment tools. These businesses must prioritize high-value extraction or specific monetization strategies that do not rely on daily habits. For B2B SaaS and productivity tools, the L7 (7-day) curve is often more relevant, reflecting the natural rhythm of the workweek. Analyzing these curves across different cohorts allows teams to visualize whether product updates or network effects are successfully shifting users toward higher frequency engagement. Beyond simple logins, the curve can be applied to core actions, such as posting content or completing transactions, to ensure the metric aligns with actual value delivery. This framework helps founders identify if a product is hitting a nerve with a core segment, even if the overall blended engagement metrics appear low.

How to build a billion dollar digital marketplace – examples from Uber, eBay, Craigslist, and more at andrewchen

Digital marketplaces often begin in small, specialized niches that are easily underestimated by traditional investors. eBay started with collectibles like stamps and coins, while Uber was initially valued based on the existing taxi market rather than its potential to expand transportation use cases. To scale these platforms into billion-dollar entities, founders must execute specific growth strategies that move beyond their initial core. The first major lever is geographic expansion. For hyperlocal marketplaces like Uber or OpenTable, building a critical mass of supply and demand in a specific neighborhood is essential. This often requires a specialized team of launchers who manage local operations, marketing, and partnerships until a city reaches self-sufficiency. Once a model works in one market, it can typically be replicated across hundreds more. Another critical strategy involves expanding product lines and price points to unlock new addressable markets. Craigslist evolved from a simple email list for San Francisco events into a global platform for jobs and housing by listening to user requests. Similarly, Airbnb uses a wide range of price points to capture different segments, from budget travelers to large families needing high-end rentals. Reducing transaction friction is the third lever. By streamlining onboarding, payments, and trust infrastructure like reviews or ETAs, marketplaces become reliable enough for high-frequency use cases. Finally, building supply and demand stickiness is achieved through market networks. This approach combines SaaS workflow tools with a marketplace, providing utility that retains users even when they are not transacting. OpenTable's seating management system is a prime example of using SaaS to anchor the supply side of a marketplace.

New data on push notifications show up to 40% CTRs, the best perform 4X better than the worst (Guest post) at andrewchen

Push notification engagement rates vary significantly by industry, with utility and financial services achieving 40% click-through rates (CTRs) compared to just 12% for e-commerce and retail. This performance gap is primarily driven by how effectively an app integrates into a user's daily routine. High-performing apps, such as Waze or Level Money, leverage notifications to provide timely, high-value information that assists with regular activities like commuting or budget management. In contrast, retail and social apps often struggle because their messages are perceived as less urgent or relevant. To bridge this gap, the data suggests three core strategies: Cadence, Personalization, and Timing. Cadence involves finding the optimal frequency for each user segment rather than a one-size-fits-all approach. Personalization requires moving beyond generic offers to provide content based on individual user interests, such as specific sports teams or shows. Crucially, this must be based on person-level data rather than device-level data to avoid sending irrelevant messages when devices are shared. Timing remains the most significant lever for optimization. Customizing delivery based on individual user preference and historical engagement patterns can result in a 384% conversion uplift. Effective push strategy transitions the notification from a disruptive interruption to a valuable service that fosters long-term retention and brand advocacy.

Why investors don’t fund dating at andrewchen

Mainstream Silicon Valley investors frequently avoid the dating category due to several structural economic hurdles. The most significant barrier is built-in churn. Unlike traditional SaaS where product improvements increase retention, a dating app that successfully matches users causes them to leave the platform. Monthly churn rates in this sector often reach 20 to 30 percent, which translates to an annual churn of over 90 percent. This requires the company to replace nearly its entire customer base every year just to maintain flat growth. Additionally, dating is a niche market with a limited shelf-life for each user, as individuals are only in the market for a specific window of time. This restricts the effectiveness of social viral loops because users often prefer privacy over inviting friends to a dating platform. Paid acquisition channels are also difficult to optimize. The low lifetime value (LTV) resulting from high churn cannot support high customer acquisition costs (CAC), a problem that traditional subscription services like Netflix do not face. Scaling is further complicated by the need for local marketplace density, necessitating expensive city-by-city expansion. Finally, a demographic mismatch exists between younger founders and older, married investors who may lack the personal context to evaluate these products. The exit landscape is also constrained, with IAC dominating acquisitions while independent IPO attempts have historically struggled.

The most common mistake when forecasting growth for new products (and how to fix it) at andrewchen

Growth is the defining characteristic of a startup, yet most growth forecasts are fundamentally flawed. The typical mistake involves multiplying a lagging indicator, such as monthly active users (MAU), by a fixed percentage to create a hockey stick curve. This approach is a vanity exercise that assumes success without explaining how it is achieved. It masks the reality that growth usually comes from a sequence of different channels such as PR, content marketing, or SEO, each of which eventually hits a ceiling. To fix this, shift the focus from outputs to inputs. A robust forecast derives the growth rate rather than assuming it. This requires identifying leading indicators specific to the business. For example, if the goal is to double SaaS revenue, the model must detail how to double the leads in the sales pipeline. This might involve increasing content production or hiring more staff. By starting with the inputs, the relationship between actions and results becomes clear. Focusing on inputs allows for the identification of operational bottlenecks. If a revenue target requires a 5X increase in sales headcount but hiring has historically been slow, the forecast is unrealistic. Similarly, if growth depends on SEO, the model must account for the time and effort required for Google to index and rank content. Ultimately, inputs are the only variables within a team's direct control. Outputs are merely the consequence of executing the plan against those inputs. The disconnect between actions and outputs in traditional models leads to a false sense of security. Entrepreneurs should assume the opposite of success and build models that highlight the specific machinery required to generate traction. This involves moving away from cookie-cutter metrics and toward a series of steps that show how scaling specific inputs results in the desired output.

Stanford CS major seeks sales/marketing monkey at andrewchen

Business Pain / Root Cause: Non-technical founders frequently struggle to recruit elite engineering talent because of a cultural perception that business roles are redundant in early-stage startups. This friction is rooted in the historical success of engineering-led companies like Facebook and HP, which suggests that technical founders can achieve significant milestones without early business intervention. Silicon Valley maintains a distinct power imbalance where engineering talent is prioritized and MBAs are often viewed as a net negative to startup valuation. Consequently, business-oriented founders must accept a secondary leverage position and focus on proving their specific utility to the venture. To overcome the difficulty of finding a technical cofounder, business leaders should select ideas where the initial hurdle is sales or distribution rather than complex engineering. Categories such as enterprise sales, marketplaces, and ad networks allow for early progress with technical-enough resources. This approach enables the business founder to build momentum and secure funding before needing to recruit high-level computer scientists for scaling. By focusing on sales-driven categories, the business founder becomes the primary driver of growth, which shifts the internal leverage. The core value a business founder provides falls into two categories. The first is operational: generating revenue, securing partnerships, fundraising, and recruiting. The second is product-oriented: design, UI/UX, copywriting, and customer development. Successful business founders are those who demonstrate expertise in these areas or take the initiative to build the first version of the product themselves. By focusing on execution rather than just the initial idea, business founders can transition from being perceived as a biz monkey to a valuable strategic partner. Demonstrating the ability to ship a good-enough version of a product is often the most effective way to earn the respect of future technical partners.

Featured Essays at andrewchen

This archive represents a comprehensive collection of essays and frameworks by Andrew Chen, focusing on the mechanics of startup growth and network effects. The content spans over a decade of practitioner-level insights, covering the transition from early-stage product/market fit to scaling via viral loops and paid acquisition. Key technical concepts include the Power User Curve for measuring engagement, the Law of Shitty Clickthroughs regarding channel decay, and the Next Feature Fallacy which warns against building features to solve underlying retention issues. The collection provides specific guidance on marketplace liquidity, emphasizing that the supply side is often the primary driver of value. It also explores the consumerization of B2B SaaS, where consumer-grade growth tactics are applied to enterprise environments. Detailed resources include investor-focused slide decks on growth accounting, frameworks for calculating Customer Acquisition Cost (CAC), and strategies for navigating the 'Trough of Sorrow' post-launch. The archive serves as a foundational knowledge base for understanding how network effects drive long-term defensibility and how to identify the red flags in growth metrics that investors prioritize during the due diligence process.