The durable profit pool often sits one layer above the business you launch with, so build the low-margin service that earns customer attention, then monetize the interface you now own.
Instacart
Instacart is an asset-light grocery-delivery marketplace: no warehouses, no fleet, no inventory, just gig "personal shoppers" fulfilling orders from partner stores. Founded in 2012 by ex-Amazon supply-chain engineer Apoorva Mehta with Max Mullen and Brandon Leonardo, its through-line is that the durable, high-margin business (retail-media advertising) turned out to sit a layer above the thin-margin logistics business it launched with, defended by a two-sided marketplace only scale can build.
Business Model: the ad engine bolted onto a delivery marketplace
The problem. Grocery delivery is a brutal business: razor-thin retailer margins, expensive last-mile labor, and capital-punishing economics if you own warehouses, trucks, and inventory. Instacart needed a model that could scale nationally without drowning in capex, and it needed a profit pool that delivery fees alone could never provide.
The approach. Stay asset-light (gig shoppers are variable labor that flexes with demand) and then monetize the customer interface it owned. Because Instacart sits between CPG brands and purchase intent at the point of sale, it built a retail-media engine: sponsored product listings, brand ads, and digital coupons sold to consumer-goods companies at very high margin.
How it solved it. In its 2023 S-1, Instacart reported $29.4 billion in gross transaction value across 263 million orders for 2022, yet advertising and other revenue already made up 29% of total revenue at roughly $740 million. Ad revenue grew about 27% to roughly $940 million in 2023 and crossed $1 billion in 2024. Delivery covers a thin logistics business while advertising, running near 75%-plus company gross margins, is where the profit actually lives, the same aggregate-then-sell-access playbook as Amazon and Google.
Network Effect: the retailer-shopper-consumer flywheel
The problem. A grocery marketplace is dead on arrival if any of its three sides is thin: too few stores means no selection, too few shoppers means slow delivery, too few customers means no reason for shoppers to work. Instacart had to make all three sides reinforce each other rather than starve.
The approach. Rather than compete with grocers, Instacart digitized incumbents, signing retail banners to expand catalog breadth, which pulled in orders, which supported more active shoppers and faster fulfillment. Density in each city became the engine: more stores lead to more orders, which lead to faster delivery and better retention, which attract more shoppers.
How it solved it. By the time of its IPO, Instacart had grown from about 165 retail partners in 2017 to over 1,400 retail banners spanning roughly 80,000 stores, representing the large majority of the US grocery market. That breadth is what let the ads business work too: 5,500 brands were advertising on a platform reaching about 7.7 million monthly active shoppers, since advertisers only show up where the demand is already aggregated.
Moats: defensibility in the data layer, not the delivery
The problem. Delivery logistics is not, by itself, defensible: routes and gig couriers can be copied by DoorDash, Uber, and Amazon, all of whom entered grocery. If Instacart's moat were only "we deliver groceries," it would be commoditized into permanently negative-to-thin margins.
The approach. Compound a proprietary data and advertising asset that only order volume can produce. Every basket teaches Instacart what shoppers buy, when, and at what price, and that first-party purchase data powers ad targeting and measurement that competitors cannot replicate without equivalent scale, turning a retail-media network into the real moat.
How it solved it. The advertising layer that grew past $1 billion in annual revenue is structurally hard to attack because it feeds on the same order flow the marketplace already generates, so scale begets better ad performance begets more advertiser spend. Instacart's own IPO framing emphasized this: the company that owns the transaction and the data, not the courier, captures the durable economics.
PMF: proving the product worked before scaling it
The problem. In 2012 it was genuinely unclear that on-demand grocery delivery via crowdsourced shoppers could work reliably enough that people would pay for it and come back. Mehta had launched roughly 20 failed startups before this, so the risk of building something nobody wanted was concrete, not theoretical.
The approach. Ship a rough product fast, prove it literally works, then obsess over operational quality metrics. To get into Y Combinator after missing the application deadline, Mehta used the Instacart app itself to deliver a six-pack of beer to a YC partner, proving the thing functioned before pitching it. The team then tracked the "found rate" (items delivered exactly as ordered), late-delivery percentage, and deliveries per week.
How it solved it. The beer stunt got Instacart into YC's summer 2012 batch and helped raise about $2.3 million in early funding, and the metric obsession kept fulfillment reliable enough to retain customers as the company expanded. The product-market signal held up: Instacart reached $29.4 billion in GTV a decade later, but the early evidence was operational, not just anecdotal.
Founder-Market Fit: an Amazon supply-chain engineer on a routing problem
The problem. Grocery delivery is fundamentally a routing, inventory, and operations problem: which shopper, which store, which items, which order, delivered on time. A founder without deep supply-chain intuition would likely underestimate exactly how hard the logistics of same-day fulfillment across thousands of SKUs really is.
The approach. The founder was purpose-built for it. Before Instacart, Apoorva Mehta spent roughly two years as a supply-chain engineer at Amazon (starting in 2008), writing systems to move goods through fulfillment networks, precisely the discipline grocery delivery demands.
How it solved it. That background let Mehta design Instacart around the operational metrics that actually govern the business (found rate, delivery timing, order density) rather than treating it as a simple app. The fit paid off concretely: Mehta took the company from a missed YC deadline in 2012 to a public listing in 2023, becoming a billionaire on the strength of a logistics business he was uniquely equipped to build.