← all systems · see it in The System
Startup company
A startup is a young company, usually founded by an entrepreneur or a small founding team, built to search for, develop and validate a scalable business model and to grow large beyond its founders (src:wp:startup-company). Management scholars describe it as a temporary organisation searching for a sustainable business model under extreme uncertainty (src:bajwa-2017-pivots, after Blank and Ries); economists describe it as experimentation: success probabilities are low, extremely skewed and unknowable until an investment is made (src:kerr-2014-experimentation). Its funding comes in stages (founders' own money and friends and family, angels, seed, Series A, B, C ... rounds), often from venture capital, and it ends by growing into an established firm (exit by IPO or acquisition) or by failing, which most do. It lives in startup ecosystems (Silicon Valley, New York, Tel Aviv) of founders, investors, universities, mentors, incubators and accelerators. Walked here as the firm together with the founders, investors and ecosystem that make it.
Properties published scienceour proposal
Sourced fields, and our own readings.
| born at | contested (C-3045): 1946 (ARDC, J.H. Whitney) for the venture-financed kind; any new firm, as old as business, for economists' 'firm startups' |
| born precision | years |
| boundary | Inside: the founders and early employees, the product being built, the startup's cash, equity and records, and the decisions taken in its name (a legal company, usually with an equity table shared among founders, employees with stock options and investors). Outside: customers and the market it is trying to find, rival startups and incumbents, investors (angels, venture capital funds and their limited partners), the labour market, the ecosystem's mentors, accelerators and universities, and the state's securities, bankruptcy and employment law. The line moves: investors hold board and control rights inside it; an exit by acquisition puts it inside another firm. |
| nested level | Built from people (founders, employees), a product, capital and contracts; a kind of business (a young firm that intends to grow large); embedded in markets, startup ecosystems and the economy. It runs on people (founders' and investors' decisions). |
| scale | social |
Measures (26) published science
| characteristic clock: exit window used to score VC-backed attempts | five years after the first VC investment |
| startup failure within five years | about 65-80% (later analyses); 90% (Fortune 2014) % of startups |
| characteristic clock: the startup process | three years or longer (one estimate) |
| VC-backed ventures shut down after 10-21 years | 75% (of 1986-1997 entrants, by 2007); matched non-VC entrants 66% |
| VC-backed startups terminated at a loss, 1985-2009 first rounds | about 55%; 6% returned more than 5x % of startups |
| share of gross VC return from the top 6% of startups | about 50% % of gross return |
| share of US firms founded each year that obtain VC | about 1,000 of over 500,000; about 40% of new US listings firms per year |
| VC opportunities considered per deal closed | about 100; median firm closes about 4 deals a year; 1.7 term sheets per close opportunities per deal |
| VC fees on limited partners' capital | 1.0-2.5% a year plus up to 20% of profits |
| cost to set up a software startup | $5 million a decade earlier; under $50,000 (2014) US$ |
| net new jobs created by firm startups, US 2005 | about 3.5 million (all private sector: about 2.5 million) jobs |
| mean founder age, top 0.1% growth firms | 45.0 (all founders hiring at least one employee: 41.9) years |
| share of startup-created jobs eliminated by exits within five years | about 40% (another passage of the same paper: about 47%) % of jobs |
| US 1996 startup cohort reaching $10 million in sales | about 3,500 of about 500,000; over half failed by 2002 firms |
| reasons given in 101 startup failures | no consumer interest 42%, cash 29%, personnel 23%, competition 19%, pricing 18% % of failures |
| time between failed attempts (success group) | decays with the number of failures n, approximately as a power law; groups distinguishable at n = 2 time |
| pivot types identified in 49 software startups | 10 types, 14 triggering factors count |
| worldwide startup network size | 41,830 companies, 117 countries, 135,099 links (1990-2015) nodes |
| degrees of separation between startups | 4.74 steps |
| venture funds' early-stage success on open deals | about 10-15%; network-centrality picks peaked at 50% (June 2003) % of deals |
| unicorns worldwide | 1,523 worth about US$5.6 trillion (January 2025); 1,248 per CB Insights (May 2024) companies |
| startups in the San Francisco Bay Area, 2024 | 14,500 (New York 12,500; Silicon Wadi about 9,000) startups |
| ARDC's return on Digital Equipment Corporation | over 1,200 times; 101% a year (1957 investment, 1968 IPO) multiple |
| Sequoia's return on Google | 320 times ($12.5 million in 1999 to over $4 billion by 2005) multiple |
| characteristic clock: spin-off generation in Silicon Valley | about 20 years for 8 founders to give 65 enterprises |
| AngelList startups funded, June 2013 | about 2,000 of about 100,000 startups; 18,000 accredited investors |
Open contentions (9) open contention
Where sources or readings disagree. Both stay on the map.
- C-413 Where do thoughts come from?
booked: No theory is proven, so the map lists the main ones side by side and doesn't pick one. It shows where thinking is handled in the brain, not where thoughts start.
the other reading: see C-413 - C-1510 Do companies get safer as they age?
booked: The map books a roughly constant death rate for listed companies, and keeps the theories that risk falls or rises with age.
the other reading: an approximately constant hazard after listing (Daepp et al., booked in sys:business) - C-2438 Do small businesses create most new jobs?
booked: The map keeps the competing findings.
the other reading: small businesses create most jobs - C-3016 At what age are founders most likely to succeed?
booked: The map keeps the competing readings; none is ruled.
the other reading: popular and investor samples: mean ages 29 (award winners), 31 (media lists), mid-30s (two VC firms' founders); Azoulay et al. note these are selected samples - C-3045 What counts as a startup?
booked: The map keeps the competing readings; none is ruled.
the other reading: economists counting all new firms call every age-0 firm a 'firm startup' (Haltiwanger et al.), which makes the startup as old as business; software-startup research: 'temporary organizations' with little operating history aiming to scale aggressively (Giardino et al.) - C-3046 Why do startups fail?
booked: The map keeps the competing readings; none is ruled.
the other reading: lack of consumer interest first (42% of 101 failures; encyclopedia); self-destruction rather than competition (Giardino et al., citing a practitioner source); neglecting the learning process in favour of building; the team (VCs name it in 92% of failures; Gompers et al.); running out of money (over a third of founders; encyclopedia) - C-3047 How many startups fail?
booked: The map keeps the competing readings; none is ruled.
the other reading: 90% (Fortune 2014, via the encyclopedia); 'more than 90%' (Giardino et al., citing a practitioner source); 75-90% for software startups (Bajwa et al., citing Nobel 2011 and Marmer et al. 2011); over half of the 1996 US cohort failed by 2002 (Shane 2009, via Kerr et al.) - C-3048 Should governments promote startups?
booked: The map keeps the competing readings; none is ruled.
the other reading: Shane: policies that encourage start-ups lead people to create marginal businesses that are more likely to fail, have little economic impact and create few jobs (via the encyclopedia) - C-3049 Can startup success be predicted?
booked: The map keeps the competing readings; none is ruled.
the other reading: early network centrality predicts success, doubling venture funds' 10-15% hit rate on open deals (Bonaventura et al.); failure dynamics separate eventual successes from the rest as early as the second failure (Yin et al.); founder personality and team mix predict success (McCarthy et al.); founder age and industry experience do (Azoulay et al.)
Cross-scale patterns seen here (24) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Startup company. Each links to every other system that shows it.
- nested ladder (structural)
Valuation rungs where each implies the one below: unicorn (over US$1 billion), decacorn (over US$10 billion), hectocorn (over US$100 billion); funding rounds climb the same way (seed, Series A with traction, B with consistent revenue, C and D toward IPO). - same role, different builds (structural)
One job, paying for a startup's early experiments before it has revenue, is done by unrelated builds: founders' savings and family (bootstrapping), angels, venture funds, equity crowdfunding, revenue-based lenders, factoring, startup studios, and established companies' internal startups. - same shape, opposite sign (structural)
One move, redrawing the product's boundary in a pivot, run both ways: zoom-in (a single feature becomes the whole product) and zoom-out (the whole product becomes one feature of a larger product), which the source defines as opposites; the platform pivot likewise runs application to platform 'or vic - trade one property for another (structural)
Financing risk is bought down with option value: taking larger chunks of money at each stage protects against not getting the next round but 'reduces the value of the abandonment options'; organisations spread along a second dial, many projects with many terminations or few with few (Nanda and Rhode - who opens the door (structural)
Capital's door into a startup is mostly opened by the investor ahead of time: most VC deal flow comes through networks and the VC's own outreach, 'Only 10% come inbound from company management'; an accelerator is a port installed in advance with fixed terms; the law opened or closed the door to publ - barrier-gates-complexity (partial)
Limited liability and soft bankruptcy are a seal between a failed venture and the founder's next one; the without-it case is shown in part: Japan's reform that 'reduced the consequences of closing a firm encouraged greater levels of entrepreneurship and risk taking'. Partial: more ventures and risk, - causal ceiling (partial)
A limit on what any investor or founder can know in advance, set by the situation, not their means: 'the knowledge required to be successful cannot be known in advance or deduced from some set of first principles'; even a top fund's partners' scores barely sort winners. Contested by studies that fin - coordination-before-mapping (partial)
Founders act before they hold a model of the market: 'not to sweat the business model too much at first'; affordable loss and building something people want come first, the business model after the learning. Partial. - holding / entrainment (partial)
A venture fund sets when its startups get money (staged against milestones), whether they die (it shuts down ventures 'receiving bad signals') and who runs them (control rights to replace CEOs). The portfolio holds its parts' clock and death; the startups are its holdings, not organs, so partial. - lightning (partial)
A discharge that changed what came after: 'The spark that set off the explosive boom of "Silicon startups"' was a 1957 dispute at Shockley Semiconductor; eight employees left, founded Fairchild, and its founders and their employees set off 'a veritable startup avalanche'. Partial: the source's own m - not-diluting wins (partial)
Ownership share as density: each round dilutes earlier holders, and early investors make their money 'by starting small and owning a larger share of the firms that turn out to be successful'; they insist on pro-rata rights and anti-dilution protection, and cheap early tests mean 'less dilution for e - partner-becomes-part (partial)
Startups partner with established firms to make their business model work, and an exit by acquisition makes the startup part of the buyer; established firms also outsource innovation to startups. Partial: the partnership-then-merger sequence is named for the kind, not traced in one case. - smooth to sharp in finite time (partial)
Tuning how much past experience is reused (k) changes the failure dynamics not gradually but discontinuously: 'the scaling exponent follows a discontinuous pattern' with critical points at k* and k*+1. Partial: a jump at a finite parameter, not a singularity in time. - symmetry breaking (name the cut) (partial)
Agents alike at the start split into two fates at a named cut: in Yin et al.'s model the critical threshold k* separates stagnation from progression, and those who eventually succeed and those who do not are 'initially similar'. Partial: shown in a model and in data, the cut a model parameter. - templating (partial)
One module reused unchanged across ecosystems: Y Combinator's 2005 accelerator model (fixed terms plus a bootcamp) copied around the world and now 'key organizations of any Startup ecosystem'; the standard sequence of rounds and terms. Partial: reuse shown, no swap test. - The carry (partial)
The wait and the lookahead: a staged startup cannot go on until the next round arrives ('financing risk ... similar to rollover risk'), so it runs at the pace of its slowest hand-off, the funding round; investors sometimes skip the wait with 'preemptive' term sheets before formal due diligence. No o - The Payment to Entropy (partial)
Two tolls on the hand-offs: money passing from limited partners through a fund to startups pays an annual 1.0-2.5% fee and up to 20% of profits, and about 55% of VC-backed startups end at a loss; jobs created by a startup cohort lose about 40% (or 47%) to exits within five years, a per-hand-off job - Umwelt: each system carves its own slice of one shared medium (partial)
A startup carves its own slice of the shared market world: nine channels (customers, capital, talent, rivals, ecosystem, technology, law, its own metrics, the founders' inner channel); the investor reading the same startup sees team first ('the management team as more important than business related - borrowed-then-held (surface)
Functions borrowed in the early phase: startup studios provide 'HR, finance and accounting, marketing, and product development'; infrastructure is rented from cloud providers; VCs hire board members and employees. The held phase (the grown firm doing these itself) is not shown in the sources read fo - outside-to-inside (surface)
The startup form taken inside established companies as internal startups at arm's length (Target began as one at Dayton's). Surface: the outside phase is not shown to come first in the sources read. - runaway from an unstable balance (surface)
Hot markets: low financing risk funds more experimental ventures, which fail more but exit at higher valuations, and investors herd around hot sectors; the internet bubble of the late 1990s. Surface: no balance point shown, only the run. - same toolkit, independent builds (surface)
Ecosystems built in many places from the same parts (research universities, technology parks, venture capital, accelerators): Silicon Valley, Boston, Berlin, Silicon Wadi, Inovallée, Trieste. Surface: the later ones are deliberate attempts to copy, the dark twin (one build inherited many times). - seed (surface)
Eight of Shockley's former employees 'gave forth 65 new enterprises' in 20 years, each going on to do the same: a small packet of people and know-how opening many firms. Surface: reproduction by spin-off, with no combinatorial unpacking shown; the 'seed round' is a name, not a sighting. - the whole sets the local reference (surface)
The period's market sets each startup's price: startups funded in active periods were 'valued higher at IPO or acquisition even compared to startups funded in less active periods exiting at the same time'. Surface: one reading of the source.
Valence: what it trades, and with whom (10) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Human (Homo sapiens) via sys:human-society · Startup company gain
- Law (rules and law) via sys:human-society · Startup company mixed
- Financial market (hierarchical tree of stocks) via sys:human-society · Startup company gain
- Market and trade via sys:human-society · Startup company mixed
- Labour market via sys:human-society · Startup company gain
- Technology via sys:human-society · Startup company gain
- Corporation via sys:human-society · Startup company mixed
- Venture capital via sys:human-society · Startup company mixed
- Startup ecosystems via sys:human-society · Startup company gain
- Entrepreneurship via sys:market · Startup company gain
Partners and parts published science
Built from / builds (13)
- Human (Homo sapiens)
- The founding team (product, marketing, finance or operations)
- The product, first as a minimum viable product
- The business model being searched for
- Equity: ownership shared among founders, employees and investors
- Employees with stock options
- Entrants and incumbents competing for a market (creative destruction)
- Ventures competing for investors' money
- Startups competing for venture capital
- Venture firms competing for the best deals
- Startups against incumbents and each other for customers
- Startups competing for skilled employees
- Entrepreneurship
Sits inside / holds (3)
expressed by (22)
- Generation/Synthesis
- Recombination/Composition
- Quality Improvement/Refinement
- Encapsulation & Modularity (Abstraction)
- Amplification/Scale-Up (Manufacture & Distribution)
- Symbiosis/Co-Creation (Collaborative Specialization)
- Evolvability/Plasticity Engineering (Design for Future Variation)
- Internal Coordination/Regulation
- Sensing/Scanning
- Hypothesis/Goal Selection
- Search/Foraging
- Uncertainty Reduction/Learning
- Risk-managed Exploration
- Active Experimentation / Probing (Optimal Experimental Design)
- Optimal Stopping & Patch-Switching
- Social Learning / Public Information Use
- Physical/Virtual Barrier
- Self-Monitoring
- Validation of Effectiveness
- Enclave/Segregation
- Triage & Prioritized Allocation
- Bridging Supply Gaps (Store & Dormancy)
floor filled by (12)
- Human (Homo sapiens)
- Law (rules and law)
- Financial market (hierarchical tree of stocks)
- Market and trade
- Technology
- Corporation
- Venture capital
- The founding team (product, marketing, finance or operations)
- The product, first as a minimum viable product
- The business model being searched for
- Startup ecosystems
- Entrepreneurship
Kinds (10)
- Business
- Entrepreneurship
- Unicorn
- Decacorn
- Hectocorn
- Internal startup
- Software startup
- Venture-capital-backed startup
- Bootstrapped startup
- Startup-studio startup
wired (2)
Notes published science
- Venture capitalists make their money on a tiny tail: of all US startups that got venture money between 1985 and 2009, about 55% ended at a loss, and the 6% that returned more than five times their cost brought in about half of everything the industry earned. Sequoia's $12.5 million in Google came back 320 times (src:kerr-2014-experimentation).
- The most successful startup founders are not twenty-somethings. Across 2.7 million US founders, the mean age of those behind the 1-in-1,000 fastest-growing new firms was 45.0, and a 50-year-old founder was 1.8 times as likely to hit the top tier as a 30-year-old; the young ages in magazine lists come from selected samples (src:azoulay-2020-founder-age).
- Failing teaches, but only past a threshold: in tens of thousands of startup careers (and in science grants), people who eventually succeed and people who never do look alike at the start, yet the eventual winners' gaps between attempts shrink as a power law, and the two groups can be told apart by the second failure. The model behind it has a phase transition (src:yin-2019-failure-dynamics).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
the boardroom (shared floors)
- L4: Measurement and validated learning: evidence from customers, own metrics and cheap tests turned into what is known
- L5: The founding team's table: founder-CEO and co-founders deciding, with investors' control rights and mentors' advice
- L6: The business model: kept (persevere) or changed (pivot) on the evidence
- L7: Purpose: grow large and exit, or close
the startup as management and economics describe it: an untested market to a scalable model, growth and exit (reader's view)
- L1: The uncertain world it searches: customers with a problem, a market that may not exist yet, rival startups and incumbents, and capital (founders' savings, friends and family, angels, venture funds)
- L2: Contact: problem and solution interviews, surveys and cold calls (market validation); pitches and funding rounds (seed, Series A, B, C ...)
- L3: Building: the product, first a minimum viable product with a fraction of the planned features, on rented infrastructure; its intellectual property
- L4: Measurement and validated learning: falsifiable hypotheses, A/B tests, key performance indicators, the build-measure-learn loop
- L5: The founding team and its decisions: founder-CEO, co-founders, heuristics and biases, persist, pivot or close; investors' control rights at the table
- L6: The business model: the repeatable, scalable way of creating and capturing value being searched for; a pivot changes it
- L7: Purpose: grow large beyond the founders and reach an exit (IPO or acquisition), or fail and close
way in: Capital: investors' money, terms and signals
- L1: Pools of capital: founders' savings and family, angels, venture capital funds (limited partners' money), crowdfunding
- L2: The round: pitch, deal funnel, term sheet, close (seed, Series A, B, C ...)
- L3: Cash in the bank and the contract: equity, convertible preferred stock, staged against milestones
- L4: Milestones reviewed: investors read the startup's results and decide to reinvest, reposition or shut down
way in: Customers: their problems and their response to the product
- L1: People and firms with a problem worth solving; early adopters
- L2: Problem and solution interviews, surveys, cold calls, email responses
- L3: The minimum viable product in customers' hands: their use and reaction
way in: Ecosystem: mentors, accelerators, advisers and networks
- L1: The ecosystem: entrepreneurs, investors, mentors, advisers, universities, incubators, accelerators
- L2: Contact: events, programmes (accelerator bootcamps), introductions through networks
- L3: Advice and connections: mentors' guidance; investors' strategic guidance and introductions to customers and hires
way in: Founders' inner channel: thought, feeling and stress
- L1: Where a founder's ideas and felt uncertainty come from (contested: see C-413)
- L2: Self-efficacy: the confidence to start
- L3: Heuristics and biases: overconfidence, illusion of control, law of small numbers, availability, escalation of commitment
- L4: Stress and exhaustion: work demands against coping
way in: Its own state: metrics, cash and usage
- L1: Its own cash and activity: money left, users, revenue (or none yet)
- L2: Continuous deployment: each release produces usage data
- L3: Key performance indicators and A/B test results
way in: Law and policy: securities, bankruptcy and employment law
- L1: Statutes: the Securities Act of 1933, the JOBS Act (2012, Title II in force 2013), bankruptcy and employment-protection laws
- L2: Rules binding the startup: whom it may solicit (accredited investors), what it must disclose
way in: Rivals: incumbents and other startups
- L1: A crowded field: incumbents and rival startups (Google entered a search market 'dominated by other players such as Yahoo! and Altavista')
- L2: The market's structure as the startup reads it: with or without a dominant design
way in: Talent: people and know-how from the labour market
- L1: Skilled people moving between firms, carrying know-how
- L2: Hiring, with stock options for buy-in
- L3: Know-how brought in: the receiving firm gains fitness
way in: Technology: what can be built and how cheaply
- L1: Enabling technology: the internet, open source software, cloud computing
- L2: Infrastructure rented in small increments
- L3: The cost of a test falls ($5 million to under $50,000 for some firms)
way out: Exit: IPO or acquisition
- L1: Shares sold to the public (IPO), or the company bought by another firm
- L7: The exit sought
way out: Hiring
- L1: Offers with stock options to people in the labour market
- L5: Founders build the team
way out: Pitches and fundraising
- L1: Investors' capital committed (or not)
- L2: A round offered, equity for money
- L5: Founders raise funds (the finance or operations co-founder)
way out: Pivots: a changed business model sent back to the market
- L1: A different customer segment or need approached
- L3: The product rebuilt (zoom-in, zoom-out, platform, technology)
- L6: The business model changed
way out: Product releases to the market
- L1: The product in early adopters' hands
- L3: Building and shipping, iteration by iteration
- L6: The business model under test
- L7: Purpose: build something people want and grow
way out: Shutdown: the startup closes
- L1: Employees released (sometimes unpaid), patents sold, founders free to restart
- L5: Founders (or investors) close it