Ocado: In Grocery, the Business Model Is the Operating Model

01

The business model

Online grocery sells convenience on the thinnest margins in retail. A typical order carries dozens of items across three temperature zones — ambient, chilled, frozen — picked and delivered within a promised window, to a customer who defects after a single bad delivery.

The record. Founded in 2000 by three former investment bankers, Ocado began supplying Waitrose customers in 2002, listed in London in 2010, and built automated fulfilment centres where robot swarms pick a fifty-item order in minutes. Its grocery technology is now licensed to major grocers across the United States, Canada, Japan, and Australia.

How the model works. The entire model is one piece of arithmetic: what an order earns against what an order costs. On the earning side, a basket of groceries carries retail’s lowest gross margins, so the average order value has to be large enough to matter. On the cost side sit the two numbers that decide everything — the cost to pick the order (labour or automation walking dozens of items into totes across temperature zones) and the cost to deliver it (a van, a driver, a time slot, the last three kilometres). Delivery cost is not fixed; it falls with density — the more orders on one van route in one area, the cheaper each becomes, which is why the same operation can be profitable in one postcode and ruinous three suburbs over. Ocado’s expression of this logic is the purest: automated fulfilment centres where robots pick orders in minutes, built on the premise that online grocery is a logistics business with a storefront, not a store with delivery attached. The same arithmetic governs every grocery and quick-commerce model anywhere, including India’s — the storefront attracts the customer; the fulfilment economics decide whether the business exists.

• Earns from: high-frequency, low-margin baskets — profit lives or dies in average order value against fulfilment cost per order. – Wins on: density and pick efficiency — orders per route, items per hour, slots filled per van. – The tension: every operational failure is public. A late van, a melted item, a bad substitution — grocery customers forgive nothing, and one poor delivery undoes months of acquisition spend.

Where this model fails. Geography first: expanding to new cities before saturating existing ones, so no route ever reaches the density where the economics turn. Basket second: chasing customer counts while average order value stays below the fulfilment cost line — growth that loses money faster at scale. Trust third: substitution and freshness failures that quietly push the highest-value weekly-shop customers back to the supermarket, leaving the model serving only the low-value convenience orders it loses money on.

02

What the case taught us

The working record stays sealed; the learning is shared.

• No gap between business model and operating model. Fulfilment design, delivery economics, and the commerce layer are not three departments — they are one machine, and a strategy that treats them separately fails precisely at the moment it scales. The categories where this holds are the ones founders underestimate most: grocery, pharmacy, fresh food, quick commerce.
• Density before geography. Grocery economics improve with order density, not coverage — ten thousand orders in one city beat ten thousand orders across ten cities by a margin that decides survival. The expansion instinct that serves most eCommerce is precisely backwards here: saturate before you spread. Most grocery failures the firm has examined were geography failures wearing marketing costumes.
• The basket is the business model. Average order value against fulfilment cost per order is the entire equation — which makes assortment strategy a logistics decision as much as a merchandising one. When a founder shows us a grocery model, the first question is never about the app. It is about the cost of the last three kilometres.

In grocery, the business model is the operating model — there is no gap between the two.

03

Chitrangana’s transformation advisory

Two forces converge on this model by 2030. Falling automation cost curves pull fulfilment economics — once the privilege of giants — toward regional players, moving the moat from owning robots to designing the operation around them. And the customer starts becoming an algorithm: household AI assistants managing the running list, reordering staples, choosing the supplier on availability, price, and reliability rather than storefront persuasion. Our advisory to grocery and quick-commerce operators, in order:

1. Fix the order arithmetic before touching anything else. Know your true fulfilment cost per order, per zone, per slot — and the average order value each customer segment must clear. Every other decision, from assortment to expansion, is downstream of this number, and most operators are running on an average that hides the losses.
2. Buy density, not geography. Direct the next year of growth spend at deepening existing delivery zones — slot utilisation, route density, basket growth within the map you already serve. Expansion earns its place only after existing zones clear their economics.
3. Become the supplier a machine would choose. Agent-driven ordering rewards exactly one profile: accurate availability, honest pricing, and reliability that is consistently true. Make your catalogue, stock position, and fulfilment promises machine-readable now — because when the household agent picks the default grocer, that choice will be nearly permanent.

Preparing the commerce layer for that buyer is the core of AI Commerce; getting the fulfilment economics right before scaling remains classic eCommerce Consulting — and in this category, the second must come first.

The next grocery winner is designed for a customer who never sees the shop.

Chitrangana