The Maze: Downloads of DoorDash, Uber Eats, Instacart and Grubhub fell an estimated 21% in the United States and Canada in 2026 to date. Sensor Tower, a market-intelligence firm, also finds that advertising supplied a larger share of those installs. That signals a tougher contest for new app users. It does not establish falling delivery orders.
The Details:
The sample leaves out major delivery channels. Sensor Tower's four-app group excludes Walmart Grocery, Amazon's Whole Foods delivery and Target's Shipt because those services operate inside broader shopping apps. Its estimated decline therefore describes installations of four specialist apps, rather than every way customers order delivery. Downloads fell 1% in 2025, making the reported 2026 decline substantially sharper. But the report does not specify its exact year-to-date cutoff or provide absolute download totals. A smaller flow of installs is a warning about that entry point; it is not a count of customers lost across the whole market.
More installs are attributed to advertising. Paid sources supplied 25.9% of the group's downloads in 2026 to date, compared with 21.0% in 2025 and 20.0% in 2024. That is a 4.9-percentage-point increase from last year. Instacart was most dependent on paid sources at 39%, up from 33%; DoorDash and Uber Eats each stood at 25%. The report does not define how it separates paid from unpaid installs. Nor does it disclose the price of an install or test how many additional downloads advertising caused. A rising paid share alone cannot establish that ads became more expensive.
Existing usage tells a different story. Combined monthly users of the four apps still grew at low single-digit rates in the report's quarterly comparisons during 2026. People can keep ordering through an app they installed years ago, so fewer downloads can coexist with continued usage. Sensor Tower's measurement service estimates downloads and active users separately using proprietary models and panel data. These figures are directional estimates, rather than company-reported order records. They cannot show how many deliveries each user bought or whether an advertising-attributed installer became a repeat customer.
MarketMaze Take: The useful finding is a weaker stream of new app installations alongside greater reliance on paid sources. Delivery operators should take that combination seriously. But treating the download decline as a demand collapse, or the paid-share rise as an ad-price index, would turn two limited measures into two unsupported conclusions.
The denominator changes the competitive picture. Excluding Walmart, Amazon and Shipt leaves important delivery routes outside the frame. For retailers and advertisers elsewhere, the portable lesson is to check which customer channels a market estimate includes before using it to judge demand. The US/Canada numbers do not establish the same trend abroad.
Advertising share needs a return measure beside it. A larger paid share could reflect fewer unpaid installs as well as changes in advertising activity. The missing cost and causal-effect evidence matters: marketing teams still need actual spending, orders and repeat purchases to judge whether acquiring those users pays. The install warning survives; a verdict on advertising profitability does not.
Sources: PPC Land, Sensor Tower
