पाठशाला Pathshala · वृद्धि Vṛddhi, Growth · Lesson 24 · Scale

Expanding to new cities: the launch playbook

A new city is a new small company with your brand on it. Choose the next five from demand you can see and economics you can compute, then launch each with a checklist and a decision date.

Pathshala, The Founder Library · 11 October 2026 · 7 min read

An aerial view of tall apartment blocks in Chennai under a clear sky. — Photograph: Daswin Ebenezer · Pexels

A company that works in Bengaluru announces ten new cities in a year. Eighteen months later it is in four of them and has quietly left six, having spent what it raised on learning that each city was a different business. The cities were not the mistake. Opening them on the same day, without a way to tell which ones were working, was.

Expansion is the most common growth plan in Indian consumer, services and marketplace businesses, because the home city fills up and the next one looks like the same opportunity again. It rarely is. This lesson gives a method for choosing the next five cities from evidence the company already holds, for computing whether each can make money, and for launching them one at a time with a checklist and a fixed date on which the company decides whether to stay.

Why a new city is a new company

For any business with a local side, supply, delivery, field staff or a marketplace, the economics in a city depend on density. Bill Gurley’s 2014 essay How to Miss By a Mile sets out why for Uber: as demand and supply both grow in a city, pickup times fall, the area served widens and drivers spend more of each hour with paying riders, which allows lower prices and brings more riders. Andrew Chen, who worked on rider growth there, describes the company as hundreds of hyperlocal marketplaces, each two-sided. None of those effects carries from one city to the next. Every new city starts thin.

The same holds for a cloud kitchen, a home-services company, a diagnostic lab, a B2B distributor or a coaching brand with centres. The brand, the product and the software travel. The density, the supply, the local prices and the reputation do not. Treat each city as a new small company with a head start, and judge it as one: does it reach the density at which its contribution per order covers its local fixed costs, and how long does that take?

Read the demand you already have

Before any city is launched, it has already told you something. Pull five signals by city for the last six months. Orders or sign-ups from outside the home city, which customers found without any local marketing. Enquiries you could not serve: the pincodes on failed delivery checks, the cities on the waitlist, the WhatsApp messages that began with can you deliver to. Search and site traffic by city from your analytics. Inbound B2B leads by city in the CRM. And repeat behaviour of the out-of-city customers you did serve, which says whether demand there is curiosity or need.

An aerial view of dense housing around a lake in Jaipur. — A city is thousands of separate neighbourhoods. The orders and enquiries already arriving show which of them want the product. Photograph: Abhinav Sharma · Pexels

Rank twenty candidate cities on these signals, scaled by population or by the size of your target segment there. The ranking usually surprises. Some large metros rank low because a competitor is entrenched; some cities in the next tier rank high because the product solves a problem there that the metros have already solved another way. The [Bharat lesson](/library/bharat-opportunity-hype-versus-paying-demand) is the caution to keep beside the ranking: interest is not paying demand, so weigh repeat orders above sign-ups and paid enquiries above clicks.

Model the unit economics city by city

For the top ten, build a one-page model each, with local numbers rather than the home city’s. Price realised, which may be lower: the same plate or service often sells for less in a smaller city. Variable cost per order: local supply, delivery distance, payment and packaging. Fixed cost per month: the city lead, the local team, rent, the warehouse or kitchen. From these, contribution per order and the orders a month needed to cover fixed cost. The [D2C](/library/d2c-unit-economics-order-that-must-make-money) and [marketplace](/library/marketplace-unit-economics-take-rate-gmv) lessons give the full structure.

A worked example. A home-services company earns ₹180 of contribution per job in Bengaluru. In a candidate city prices are 15 per cent lower and technician wages 20 per cent lower, so contribution is about ₹150. The city needs a lead, a small operations team and an office, about ₹6 lakh a month, so it breaks even at four thousand jobs a month, or about 130 a day. The demand signals show three hundred unserved enquiries a month from that city today. Getting from three hundred to four thousand is the real question, and the model makes it visible before a rupee is spent. Add the one-off launch cost, typically the first three to six months of fixed cost plus launch marketing, and the cash each city consumes before it breaks even.

Two costs are easy to leave out of the model and usually decide it. The city lead. A good person who knows the city, its suppliers and its customers will cost more than the model’s first guess and is worth it; a manager sent from head office for six months learns the city at the company’s expense. The distance from the hub. Every city that cannot share a warehouse, a kitchen, a training centre or a regional manager with an existing one carries those costs alone. A city three hours by road from the home city and one that needs a flight are different propositions even when their demand signals match.

Choose five and put them in order

Score the ten on four columns: demand signal, contribution per order, the gap between today’s demand and break-even, and how easily supply can be built there. Gurley’s marketplace checklist, All Markets Are Not Created Equal, is a useful prompt for the last column: he notes that once a process model is established, new city launches for companies like Yelp, Uber and GrubHub were relatively quick, and that the difficulty of signing up suppliers decides how fast a rollout can go. Take the top five. Put them in an order that lets each launch help the next: a cluster near the home city first, sharing supply, warehouses and managers, before a distant metro.

Then launch one at a time, or two at most. Paul Graham’s Do Things that Don’t Scale describes Facebook starting only at Harvard, and the principle he draws from it: a deliberately narrow market is like keeping a fire contained at first to get it really hot before adding more logs. The same essay notes that Airbnb’s early work included going door to door in New York. Within a city, start in a catchment of a few pincodes and fill it before widening. Density inside a small area beats coverage across a large one.

Open a city on the day you can name the date you will decide whether to stay.

The launch and the review date

The checklist above is the playbook, and it should get better with every city. Three items carry most of its weight. Supply before demand: a city that markets before it can serve spends money on a reputation for failing. City-level data from day one: every order, cost and lead tagged with the city and pincode in your own database, so the review is about numbers. A review date set at launch, ninety to one hundred and twenty days out, with three thresholds written down that day: weekly orders, contribution per order and the repeat rate of the first cohorts. On the date the city lead and the founders take one of three decisions: invest more, hold and fix one named thing by a second date, or close.

Closing a city is part of the playbook. Write down what the city taught, which assumption in the model was wrong and by how much, and correct the model for the next five. Companies that cannot close a city keep paying for the ones that are not working out of the margin of the ones that are, and the whole network drifts towards the average of its weakest markets.

Read the first cohorts, not the launch week. Launch weeks are flattered by the waitlist, by curiosity and by launch offers. The question the review answers is whether customers who arrived in the second and third months, after the offers ended, came back at a rate close to the home city’s. If they did, a city that is behind on orders is a marketing problem and can be fixed with spend. If they did not, it is a product or service problem in that city, and more spend will only buy more customers who leave.

The monthly expansion review

On the first working day of each month, one row per live city: weekly orders, contribution per order, fixed cost, months to break-even at the current growth rate, repeat rate of the latest cohort, and days until its review date. One row per candidate city: the latest demand signals and any change in the ranking. Open the next city only when the last one has passed its review, or two in a row have. Once a quarter, compare each city’s actual path with the model you wrote before launch, and correct the model where it was wrong. The model is the asset; the cities are the evidence that improves it.

The rupee figures are illustrations. Check local registrations and tax with your CA before opening in a new state; nothing here is legal or tax advice.