The tests to run before a marketplace opens its next city or category

Expand when the home market keeps its service quality after incentives come down, a new manager can run the launch playbook without the founders, and the team knows whether the channel that built demand starts warm or cold in the next market. Then run market two as an experiment with its bars and its team written down first.

• Pull the subsidies and see what holds. Adriel Frederick's main test from Lyft is whether pickup times, reliability and prices hold once driver incentives and rider discounts come down. He would look at the tail, meaning bad hours, neighborhoods and weather, and wants utilization inside a band, with drivers busy enough to keep logging on and enough slack that a rider in a quiet area still gets a car at a normal price. He would also want positive contribution per transaction at normal pricing before expanding.
• Expand only once the launch is written down. Casey Winters says Grubhub ran only in Lincoln Park in Chicago for a while, and its launches followed a written playbook: two salespeople with a contract quota, then live listings, online ordering, AdWords and SEO pages. If a team can't write down which actions unlock liquidity and who owns each one, he would launch one market at a time. His checklist ends with the second market reaching liquidity on the playbook without founders running it.
• Rank cities by demand that shows up unprompted. Gustaf Alströmer would track one metric that needs both sides, such as trips or nights booked, and check whether it keeps growing with the ads off. He would rank targets by searches, signups and waitlist entries from outside the core market. Later he adds the cost side, including the regulatory fight Adriel raises and local product changes such as payments or ID checks, and puts each candidate city on one sheet.
• Check whether the channel and its pages carry over. Luc Levesque would name the one channel that made market one work and check whether it starts warm or cold in the next market. Domain authority travels, he says, but page content doesn't, and a city page with no reviews ranks badly and converts worse. Casey's answer is that Thumbtack and early Grubhub filled pages before supply signed up, with lead forms routed by hand and menus collected from every restaurant they could find.
• Treat a new category as a possible new company. Brian Balfour is more cautious about categories than cities. Two orders a day per restaurant says nothing about whether florists or caterers will stick, so before launching he would check that the core use case and supply behavior stay the same and that market one's channel works for the new category. If the supply behaves differently, Gustaf would score that category as a new company.
• Write down market two's bars and staffing before launch. Adriel would run market two as an experiment, with its criteria and the response to a miss decided in advance, after a Lyft pricing system his team held onto too long. Brian adds the opposite failure, killing market two after under-resourcing it, for example by sharing a team with market one. He would name a dedicated team and budget next to the bars, and add resources quickly once the early bars are hit.