What founders, investors and authors say works — summarised in plain words, linked to
the source, and tried on the businesses we run. When enough of them have tried it, we say whether it worked.
121 classics · all free0 being tested0 with a verdict5 businesses before a verdict
Networks and marketplaces
Build the atomic network first (constrain the market)
Andrew Chen: Led rider growth at Uber; now a partner at venture firm Andreessen Horowitz.
“the “atomic network” is the smallest network needed that can stand on its own”
Before going after a big market, make one tiny group (a single campus, one team, one busy street corner at rush hour) dense enough to keep working without help, then copy that group next door. Facebook opened only to Harvard email addresses and Uber began in San Francisco; in Lenny Rachitsky's study, all but one of 17 marketplaces first limited themselves by place or by category.
When it applies: The product is only useful once enough of the right people are on it in the same place, category or moment.
Not tried yet
None of our businesses has tried it yet.
Solve a hard problem for the hard side of the network
Andrew Chen: Led rider growth at Uber; now a partner at venture firm Andreessen Horowitz.
“The Hard Side of a network is, by definition, hard to scale.”
In most networks a small minority (sellers, creators, drivers, hosts) does most of the work and holds most of the power. Early product effort should go to a real problem those people have, because the rest of the network follows them.
When it applies: One group of participants is scarce, does more work than the others, and can easily leave.
Chris Dixon: Andreessen Horowitz partner leading its crypto funds; sold his startup Hunch to eBay.
“The tool helps get to initial critical mass. The network creates the long term value for users, and defensibility for the company.”
Give the first users something useful they can do on their own, so the product is worth using before anyone else is there; once enough people use that tool, add ways for them to share, find each other or trade. The solo tool gets you started and the network is what keeps people.
When it applies: The network is empty at launch but at least one side could get value from the product alone.
Not tried yet
None of our businesses has tried it yet.
Geoffrey G. Parker: Dartmouth engineering professor and a pioneer of the economics of platforms.
Marshall W. Van Alstyne: Boston University professor and a pioneer of the economics of platforms.
When each side of a market values the other, it can pay to give one side the product cheaply or free, even permanently, and make the money on the other side. In practice founders subsidise the scarce side whose presence draws the other in, for example by guaranteeing early suppliers a minimum income or waiving their fees.
When it applies: One side is scarce or price-sensitive and its presence is what persuades the other side to pay.
Contested. Subsidies spend real money (on this platform, real wallet dollars) and can attract participants who leave when payments stop. Guaranteed-earnings schemes for individual workers can raise employment-classification questions, and pricing below cost to push out a rival can raise competition-law (predatory pricing) concerns. A subsidy can also hide whether either side would pay full price.
Not tried yet
None of our businesses has tried it yet.
Lenny Rachitsky: Former Airbnb product lead; writes Lenny's Newsletter, one of Substack's largest.
Most successful marketplaces in Rachitsky's study (14 of 17) put nearly all early effort into getting sellers or providers, because supply tends to bring its own buyers. They mostly did it unglamorously: one-to-one sales and outreach was the most common lever (about 60% of companies), then referrals and piggy-backing on existing networks, and the typical company leaned on just two levers.
When it applies: An early marketplace where buyers will not come until there is enough to choose from.
Sarah Tavel: Benchmark venture partner; earlier Pinterest's first product manager.
“It is far more about making sure you onboard that demand side, because if you have demand then supply will come.”
In business-to-business marketplaces the buyers are the side to win first: suppliers will put up with an imperfect platform to reach real orders, whereas a pile of easy-to-join suppliers with no buyers goes nowhere.
When it applies: A B2B marketplace where suppliers are plentiful and hungry for orders but buyers are few, concentrated or hard to win.
Geoffrey G. Parker: Dartmouth engineering professor and a pioneer of the economics of platforms.
Marshall W. Van Alstyne: Boston University professor and a pioneer of the economics of platforms.
A business that connects outside producers with consumers plays by different rules from one that makes a product and pushes it down a pipeline to buyers. Platforms win by bringing in and coordinating outside participants and growing the value of the whole ecosystem, not by owning resources and tuning internal operations.
When it applies: Deciding whether to make everything yourself or let outside producers supply and sell through your product.
Can't test here
Moving from making the product to hosting outside producers is a business-model change whose payoff depends on building an ecosystem over years.
Metcalfe's law: network value grows with the square of users
Bob Metcalfe: Co-invented Ethernet, founded 3Com, and won computing's top prize, the Turing Award.
The value of a communications network grows roughly with the square of its connected users, because each new user can connect with every existing one; in 2013 Metcalfe fitted the rule to Facebook's user and revenue history. Critics (Briscoe, Odlyzko and Tilly, 2006) argue most possible connections are worth little, so value grows far more slowly, closer to n times log n.
When it applies: Estimating how much more valuable a network becomes as it grows, or justifying spending to add users.
Contested. Treating n-squared as a planning rule overstates what each extra user is worth and can justify overspending on growth or overpaying for networks; the 2006 critics link that optimism to dot-com era valuations.
Can't test here
It is a claim about how value scales with network size, a curve a small business cannot observe at its scale within months.
Seed the empty network yourself (Reddit's founder-made accounts)
Steve Huffman: Co-founded Reddit and leads it as CEO.
Alexis Ohanian: Co-founded Reddit and the venture firm Seven Seven Six.
A new community looks dead when it is empty, so its founders can supply the first content themselves. At Reddit, Huffman and Ohanian posted the kind of links they wanted to see through many invented accounts for the first couple of months, which set the tone and made the site look busy; the version worth testing is the open one, where founders, staff or clearly labelled house accounts seed content until real users take over.
When it applies: A community, content or listings site that new visitors judge by how active it already looks.
Contested. Posting under invented identities deceives users; if found out it damages trust, and passing off fake activity as real can be a deceptive practice under consumer-protection law and break platform terms. In the US the FTC's 2024 rule on reviews and testimonials bans reviews attributed to people who do not exist and requires company insiders to disclose their connection. Only the disclosed version (named staff or labelled house accounts, never fake reviews, buyers or sellers) belongs on this platform. Seeded content can also hide whether real users want the product at all.
Not tried yet
None of our businesses has tried it yet.