OpenTable

OpenTable is the canonical "sell the tool first, build the network second" cold-start story: it handed restaurants back-office software that was useful with zero diners, then converted that installed base into a two-sided reservations network that became the default way Americans book a table. The through-line: the same Electronic Reservation Book that solved the supply side's paperwork quietly generated the live availability data the consumer marketplace needed, and once a city crossed local density the network fed on itself.

Cold Start: Leading with the Electronic Reservation Book instead of a consumer marketplace

The problem. When Chuck Templeton, Sid Gorham, and Eric Moe founded OpenTable in San Francisco on July 2, 1998, they hit a hard chicken-and-egg trap: diners would not visit a reservations site with no restaurants, and restaurants would not list for a site with no diners. Worse, most restaurants in 1998 still ran reservations on a paper book, and many host stands lacked internet or even power, so there was no digital availability to plug a consumer site into even if diners showed up. The supply side did not exist in usable form.

The approach. Rather than launch a marketplace and beg restaurants to join, OpenTable led with the Electronic Reservation Book (ERB), a touchscreen terminal that replaced the paper book at the host stand and improved the restaurant's operations on day one, with no diners present: cleaner table management, no double-bookings, guest history. Restaurants paid roughly a $500 install fee (which often cost OpenTable around $5,000 to deliver, running cables through walls and basements) plus a $199 monthly subscription. The product justified itself as a standalone tool before any network existed.

How it solved it. The ERB locked in supply that did not depend on demand, and every terminal fed real-time availability into one database, so OpenTable built the inventory the consumer site would eventually need. After nearly running out of money, the company halted broad marketing and narrowed to four dense cities (Chicago, New York, San Francisco, and Washington, D.C.), selling door-to-door to high-end restaurants. That sequencing (win supply to local density, then turn on diners) is why supply was never starved waiting for demand: restaurants were paying for software they wanted regardless.

Network Effect: Local density flipping the flywheel and the ERB creating switching costs

The problem. Reservations are inherently local: a diner in San Francisco does not care about tables in Chicago, so OpenTable could not win once nationally. It had to reach critical density market by market, and until a neighborhood had enough bookable restaurants that a diner could reliably find a table, the consumer side had no reason to return.

The approach. OpenTable made diners free and monetized restaurants, charging a per-seated-cover fee (around $1 per diner) on top of the subscription, so it profited only when it delivered incremental, high-intent diners into otherwise empty tables. Once a city crossed density, the two-sided flywheel turned: more bookable restaurants drew more diner traffic, which delivered more covers, which pulled more restaurants onto the ERB to capture that demand, deepening inventory. The ERB itself created hard switching costs, since host-stand staff were trained on it and guest data lived inside it.

How it solved it. The strategy compounded into scale: OpenTable IPO'd on NASDAQ under ticker OPEN on May 21, 2009, one of the first tech IPOs after the financial crisis, netting about $55.8 million. By the time Priceline (Booking Holdings) acquired it on June 13, 2014 for $2.6 billion ($103 per share, all cash), OpenTable was seating more than 15 million diners per month across more than 31,000 restaurants. Ripping out the ERB meant going dark on that diner demand, which made the network durable for years.