Step 3 of 4 · 2020–26

Use what shoppers find to make the next basket more reliable

The pandemic gave this network an extraordinary volume of work. Gross transaction value rose from $5.1 billion in 2019 to $20.7 billion in 2020. That surge was a category-wide demand shock, and it exposed a weakness in using stores Instacart did not control: an item shown online might be gone when a shopper reached the shelf. Instacart’s engineers describe inventory volatility, canceled orders and the need to bring hundreds of thousands of new shoppers onto the service. Selling access to local assortment required getting that assortment right often enough for customers to return IC19 IC7.

Each completed shopping trip could help the next one. Found and not-found reports supplied fresh signals for availability models; the product showed stock badges and let customers request scarce items. Engineers reported fewer baskets containing low-availability goods and more orders after those changes. Their research also found that customers whose orders had a perfect found rate were twice as likely to return and reorder as those with a low found rate. Those observations explain why shared inventory knowledge mattered beyond route optimization IC7.

The pandemic’s growth rate did not last: orders were nearly flat in the first half of 2023. By 2025 annual orders had risen to 338.8 million, and the company reported a five-percentage-point improvement in its perfect order fill rate, which includes successful replacements. That broader measure differs from finding every original item. Repeat use depended on the service continuing to work after emergency demand subsided IC1 IC12.

Rivals Amazon entered this period with Whole Foods and Prime delivery, after buying the chain in 2017 and folding Fresh delivery into Prime in 2019. Instacart served separately managed grocers whose shelf availability it had to predict from shopper observations IC10 IC11 IC7.

Data network effects

Step 4 of 4 · 2018–26

Reuse the shopping network to sell ads across retailers

Shopping activity also supplied something brands would pay for: a chance to reach a customer choosing what to put in the basket and measure the resulting purchase. Advertising was already part of Mehta’s four-party description of the business in 2018. In 2022, advertising and other revenue reached $740 million, or 29% of revenue. This added a second way to earn from the transactions and retailer relationships the delivery operation had assembled IC3 IC1.

Jan-Erik Asplund’s Sacra analysis identifies this move toward retailer software and advertising as the next stage of the business. Carrot Ads extended the logic beyond Instacart’s own destination: retailers could use its advertising technology and advertiser demand on their sites, with revenue-sharing arrangements. In April 2025 Uber said it would use Carrot Ads for US grocery and retail advertising, connecting its marketplace to demand from more than 7,000 brands; more than 220 retailer banners already used Carrot Ads. A grocery delivery rival could also become a customer for the advertising infrastructure IC18 IC6 IC14.

The economic advantage comes from reusing access and infrastructure. An advertising-only entrant would need grocery traffic, purchase measurement and retailer access; a small delivery entrant would need to recruit advertisers and build campaign tools. Instacart could offer both through relationships and technology it already operated. The benefit reached commercial scale: in 2025 it reported $1.065 billion in advertising and other revenue, $3.742 billion in total revenue and $447 million in GAAP net income. In the second quarter of 2026 it processed 90.3 million orders and reported $111 million in net income IC1 IC14 IC12 IC13.

The advantage remained tied to retailer cooperation and the affordability of delivery. Ross’s criticism shows why grocers wanted to preserve their own identity; Instacart’s 2025 letter says affordability remained a barrier and that grocers offering in-store price parity grew faster than the platform overall. Amazon could fund grocery delivery from its broader Prime relationship. Instacart’s defense was the ability to sell a shared grocery and advertising system across other retailers while continuing to improve the shopper’s purchase IC15 IC12 IC11.

Rivals Amazon combined grocery delivery with Prime and its own stores. Instacart built a shared system for other grocers and extended its ads infrastructure even to Uber Eats. Both could monetize shopping intent; Instacart’s particular advantage was reusing its retailer network across marketplace, storefront and advertising sales IC11 IC14 IC1.

Scope economiesMulti-Product