Start with the narrowest possible slice, one place and one topic, keep friction near zero, and let each new category and location stack supply and demand until the network is undislodgeable.
Craigslist
Craigslist is a classified-ads platform that grew from one man's email list into one of the most-visited sites on the web, run for decades by a tiny staff. Its through-line: pick the narrowest possible beachhead, one city and one topic, keep the friction near zero, and let each new category and each new city stack supply and demand on top of the last until the network becomes impossible to dislodge.
Beachhead: one email list of San Francisco events
The problem. In 1995 Craig Newmark had no marketplace, no company, and no plan to build one. He simply wanted to share notices of interesting events with fellow software and internet developers around the San Francisco Bay Area. The question that faces every marketplace, how to start when you have neither buyers nor sellers, was answered by not trying to build a marketplace at all.
The approach. Newmark started with the smallest viable slice: a personal CC email list to friends, covering local Bay Area events. When friends suggested a name, he wanted "sf-events," but they pushed for "craigslist" to keep it personal and human. He deliberately did not moderate or broaden it beyond that one city and that one use.
How it solved it. The narrow focus let the list find product-market fit in one place before scaling. By mid-1995 it had grown to 240 addresses, past the limit Newmark could fit in a CC field, forcing the move to a list server, and it became a web-based service in 1996. Only after users themselves started posting jobs and items did new categories get added, so demand pulled the expansion rather than Newmark guessing at it.
Network Effect: stacking cities and categories onto a proven model
The problem. A classifieds site is only useful where it already has density: nobody posts an apartment in a city with no renters browsing, and nobody browses a city with no listings. Winning San Francisco did not automatically make Craigslist useful in Chicago or New York, and each new market restarted the chicken-and-egg problem from scratch.
The approach. Craigslist expanded deliberately along two axes at once, geography and categories, while keeping nearly everything free to minimize friction. In August 2000 it added seven major U.S. metros in one move (Chicago, Los Angeles, New York, Portland, San Diego, Seattle, and Washington, D.C.), followed by Sacramento in October, prioritizing large cities where supply and demand would concentrate fastest. It monetized only a sliver, paid job postings in a handful of cities, leaving categories like for-sale, housing, and community free.
How it solved it. Concentrating on big metros let each market cross the density threshold quickly, and every added category (jobs, apartments, for-sale, personals) gave the same users more reasons to return, deepening stickiness on both sides. The result compounded into a durable network: Craigslist now runs roughly 700 cities across about 70 countries, and revenue that peaked around $1 billion in 2018 was still in the hundreds of millions years later, all sustained by a famously small staff and a bare-bones site that barely changed.