Historical caseEvent / 2024
Unity's Runtime Fee reversal: platform dependency is a planning cost
A 2023–24 policy case about what happens when a creator platform changes the economic terms around a shipped product.
3 min read · 4 sources
36 CASE-STUDY ENTRIES / SEPTEMBER 2026 RESEARCH
How companies got built.
Different starting positions. Different capital structures. Different outcomes. Read the conditions around the decisions before borrowing the lessons.
Read the full chapter & its limitationsHISTORICAL BACKFILL / PREPARED SEPTEMBER 2026
Six more decisions worth studying.
Prepared 16 September 2026. New source checks and historical backfill, separate from the supplied research snapshot.
Historical caseEvent / 2024
Raspberry Pi's 2024 IPO: distribution is part of the hardware product
A prospectus case on how a low-cost computing brand paired community demand with OEM, reseller, and licensing channels.
2 min read · 5 sources
Historical caseEvent / 2024
iRobot and Amazon: what a terminated deal leaves behind
A January 2024 case in regulatory risk, break fees, and the operating shock that can follow an unrealized exit.
3 min read · 4 sources
Historical caseEvent / 2021
GitLab's 2021 S-1: the economics behind a developer buyer
A public filing case about converting an open-source adoption path into an enterprise expansion business.
3 min read · 4 sources
Historical caseEvent / 2018
GitHub and Microsoft: buying a developer rendezvous point
A 2018 acquisition case about preserving a trusted platform while routing enterprise distribution through a much larger owner.
3 min read · 2 sources
Historical caseEvent / 2015
Etsy's 2015 IPO: marketplace operations behind the listings
A filing-driven case on seller services, mobile conversion, and the practical work of turning marketplace scale into revenue.
3 min read · 3 sources
36 entries
Zoho: thirty years, no outside capital, $2B scale
In 1996, network-equipment vendors needed software to manage and test their gear. Sridhar Vembu, his brother Kumar, and Tony Thomas founded AdventNet in Pleasanton, California, initially as a contract software shop building network-management components for OEMs like Cisco. The "problem" they solved first was not a…
Read the case37signals / Basecamp: the deliberately small company
In 1999, Jason Fried ran a small Chicago web design consultancy called 37signals. By 2003 the firm needed a way to keep client projects organized — status, files, messages, deadlines — because email threads were losing information. They built an internal tool.
Read the caseMailchimp: twelve years of a side project, then a $12B exit
Ben Chestnut and Dan Kurzius ran a web design agency, Rocket Science Group, in Atlanta. Clients kept asking for email newsletters. The agency built a tool to do it, and treated it as a sideline for years.
Read the caseKit (formerly ConvertKit): the near-failure that a public teardown saved
Professional bloggers and course creators needed email marketing built around people rather than lists — tagging, automation, and selling — and the available tools (Mailchimp, AWeber) were built for a different job.
Read the casePlausible Analytics: the deliberately modest outcome
Included specifically because most case-study collections have nothing at this scale, and this scale is where most successful bootstrapped companies actually live.
Read the casePieter Levels: the one-person portfolio, and the seventy failures
Levels, a Dutch developer, was travelling and working remotely in 2014 and wanted to know which cities had good internet, low cost of living and decent weather. No one had assembled that data.
Read the caseGumroad: a venture-backed company that failed at being venture-backed and succeeded at being small
This is the single most instructive case in this chapter, because the founder documented the failure in real time and in public.
Read the caseTinyPilot: a $598,000 exit, documented line by line
Included because small exits are the modal successful outcome for bootstrapped software and hardware businesses, and almost nobody writes them up honestly.
Read the caseFigma: ten years of R&D, a dead $20B acquisition, and an 80% drawdown
Design tools in 2012 were single-player desktop applications. Files lived on individual machines, collaboration meant emailing versions, and developers could not see designs without the designer exporting them. Meanwhile Adobe owned the category with software built for print-era workflows.
Read the caseNotion: nearly dead in 2015, then a decade of compounding
Ivan Zhao's thesis — heavily influenced by Douglas Engelbart and Alan Kay — was that computers should let ordinary people build their own software, and that the app-per-task model (one tool for docs, one for tasks, one for databases) was a failure of imagination.
Read the caseSlack: the most famous pivot, and why it is badly misread
The original problem — the one they were actually solving. There wasn't one. Slack was an internal tool.
Read the caseShopify: a snowboard shop and a twenty-year compounding curve
In 2004 Tobias Lütke, a German programmer who had moved to Ottawa, tried to build an online store to sell snowboards with his partners (Snowdevil). The available e-commerce software — Miva, osCommerce, Yahoo Stores — was bad enough that he decided to write his own.
Read the caseAirbnb: the rejection period, and what the cereal story actually was
In October 2007, Brian Chesky and Joe Gebbia could not make rent on their San Francisco apartment. A design conference was in town and hotels were full. They put three airbeds on the floor and built a one-page site.
Read the caseCanva: a hundred rejections, a wedge product, and a $42B private valuation
Melanie Perkins, teaching design software at the University of Western Australia in Perth around 2007, observed that students took a full semester to learn the basics of Adobe's tools. The software was expensive, complicated, and desktop-bound. Her insight was not "design should be easier" — it was that the learning…
Read the caseBumble: a $13B IPO, a 97% decline, and a founder's return that did not work
This is the most important case in this section, because the standard version stops at the IPO.
Read the case23andMe: from $6B SPAC to Chapter 11 to a nonprofit owned by its founder
Anne Wojcicki's thesis, from 2006, was that consumer genetic data at scale would be more valuable than any pharmaceutical company's private dataset, and that consumers would pay for the sequencing that built it. The business model was always: sell tests to fund a database, then monetize the database through drug…
Read the caseStitch Fix: a real IPO, a 95% drawdown, and a slow, unglamorous recovery
Katrina Lake, at Harvard Business School in 2011, believed apparel retail could be personalized: rather than browsing, a customer would receive a curated box of five items selected by a human stylist informed by data, keep what they wanted, and return the rest.
Read the caseCalendly: an immigrant founder, a drained 401(k), and a viral loop
Scheduling a meeting takes five emails. The back-and-forth is pure coordination overhead with zero value created.
Read the casePartake Foods: the CPG path, which is nothing like the software path
Included because consumer packaged goods is where a large share of underrepresented founders actually build, and its economics are completely different from everything else in this chapter.
Read the caseFlutterwave: Africa's biggest fintech, and its governance problem
Included because the governance story is inseparable from the growth story, and most coverage of African startups reports one without the other.
Read the caseNubank: regulatory arbitrage at continental scale
Brazilian retail banking in 2013 was an oligopoly of five banks with roughly 80% of the market, credit card interest rates that could exceed 400% annually, branch visits required for basic tasks, and annual fees on everything.
Read the caseKlarna: a $45.6B mark, an 85% writedown, and a recovery that the market has not rewarded
In 2005 three Stockholm School of Economics students — Sebastian Siemiatkowski, Niklas Adalberth and Victor Jacobsson — entered a business-plan competition with an idea for invoice-based payment: let shoppers receive goods first and pay later, with the merchant paid immediately and the risk sitting with the…
Read the casePaystack: a $200M acquisition, and what a good non-US exit looks like
Included as a contrast to the "IPO or bust" framing, and because acquisition is the realistic exit for most non-U.S. companies.
Read the caseCommonwealth Fusion Systems: the best-capitalized bet on a thing that has never worked
Fusion power has been "thirty years away" since the 1950s. The specific bottleneck for tokamaks is magnetic field strength: confinement scales steeply with field, and conventional superconducting magnets cap out.
Read the caseZipline: ten years of operating in Rwanda before anyone in the U.S. cared
Blood products, vaccines and emergency medicines have short shelf lives and unpredictable demand. In countries with poor roads, rural clinics either overstock and waste, or understock and patients die.
Read the caseGinkgo Bioworks: what a $15B SPAC looks like five years later
Ginkgo's pitch, from MIT synthetic-biology researchers led by Jason Kelly and Tom Knight (founded 2008), was that organism engineering should become an industrial platform: a "foundry" with massive automation and scale economics that would make designing a microbe dramatically cheaper, with Ginkgo taking cash plus…
Read the caseMidjourney: roughly $500M of revenue, forty people, no investors
Text-to-image generation was a research capability in 2021 with no consumer product around it.
Read the caseHugging Face: a failed teen chatbot that became AI infrastructure and sold for $12.9B
The original product, which failed. Hugging Face launched in 2016 as a chatbot app for bored teenagers — an emotionally-aware AI friend, named after the 🤗 emoji. Clément Delangue, Julien Chaumond and Thomas Wolf raised seed money for it. It did not work as a consumer business.
Read the caseAnysphere / Cursor: the fastest revenue ramp on record, and the dependency underneath it
LLMs could write code; existing editors treated them as an autocomplete sidebar rather than rebuilding the editing experience around them.
Read the caseQuibi: $1.75B raised, 6 months live, no validation at any point
What it was. A mobile-only streaming service with professionally-produced 5–10 minute episodes ("quick bites"), with a patented "Turnstyle" feature that reformatted video between portrait and landscape.
Read the caseConvoy: a $3.8B "Uber for trucking" that was really a low-margin brokerage
What it was. A digital freight brokerage matching shippers with carriers via an app, founded 2015 by Dan Lewis and Grant Goodale, both ex-Amazon.
Read the caseIRL: $170M raised, 12 million "users," and federal fraud charges
What it was. A group-messaging and events app founded by Abraham Shafi, which reached a $1.17B valuation in a SoftBank-led $170M Series C in June 2021.
Read the caseBench and Olive AI: two ways for a services business to die
Bench Accounting (Vancouver, founded 2012) offered bookkeeping for small businesses: software plus human bookkeepers, on subscription. It raised roughly $113M, served an estimated 35,000 small businesses, and abruptly shut down on December 27, 2024 — between Christmas and New Year, with clients losing access to their…
Read the caseTheranos and Nikola: the two shapes of startup fraud
These belong together because they illustrate the same mechanism with different physics.
Read the caseWeWork: the largest single destruction of venture capital
What it was. A commercial real estate arbitrage — take long leases, subdivide, sublet short-term — marketed as a technology company that would "elevate the world's consciousness."
Read the caseThe second-act arcs: what actually transfers from failure
Rather than four separate case studies, this is the pattern across four well-documented pivots, because the pattern is the lesson.
Read the caseNo entries match. Try a different term or clear the filters.