Why Advantage+ Shopping campaigns plateaued in Q1 2026
Advantage+ Shopping was the breakout Meta product of 2024 and the consistent CPA-killer of 2025. Q1 2026 was the first quarter most accounts stopped seeing improvement. The reasons are structural.
Advantage+ Shopping was the breakout Meta product of 2024 and the consistent CPA-killer of 2025. Q1 2026 was the first quarter most accounts stopped seeing improvement. CPAs flattened. ROAS held but did not climb. The campaigns that compounded for 18 months started behaving like every other paid-acquisition surface.
Three structural reasons explain the plateau. None are obvious from inside Ads Manager.
Creative saturation. ASC's edge for most of 2025 was the model's ability to surface high-performing creative combinations to audiences detailed targeting would have missed. By late 2025, most accounts had run their creative library through enough variations that the marginal lift from another rotation dropped close to zero. The model can only match what is in the library. Stop expanding the library, stop finding winners.
Audience exhaustion inside the broad reach. ASC defaults to maximum geographic and demographic breadth, with the algorithm narrowing dynamically. After a year of compounding spend, the high-converting cohorts inside that broad reach have already been worked. Frequency mostly absorbed by people who have either already converted or definitively will not. Net-new reach inside the same geo-demo envelope is now scarce. The Q1 2026 retention extension to 730 days compounds the problem by keeping those exhausted cohorts visible inside the audience for four times longer than the architecture assumed.
The AI Overview spillover. Meta does not run on Google. Meta's prospecting layer relies on commercial-intent signals that increasingly resolve inside an AI summary rather than a click. A buyer who used to search "best running shoes under $200", click three competitor sites, and then see the retargeting ad now reads the AI Overview and never enters the click-and-cookie funnel. The signal feeding Meta's audience-building infrastructure has thinned at the top.
What separated the accounts that broke through
Three patterns recur across the accounts that kept improving CPAs through Q1.
They refreshed creative on a 14-day cadence instead of a 30-day one. The library expansion rate was the binding constraint, and they treated it that way. Five to seven new creative concepts per fortnight, each tested in the same Advantage+ Shopping campaign rather than in isolated A/B campaigns.
They added 4-5 catalog-set segmentations to the ASC structure rather than running one giant catalog. Same creative inputs, but the catalog-set granularity gave the model more lever points to optimise against. The Pixel and CAPI Auditor checks flagged catalog-event misalignment in roughly half the accounts that audited. Cleaning that up before adding segmentation produced the cleanest lift.
They moved away from "max reach" prospecting and toward existing-customer budget caps. ASC's March 2026 update added explicit existing-customer budget controls (Meta Ads platform updates, March 2026). The accounts that capped existing-customer spend at 20-30% of the ASC budget redirected the remaining 70-80% toward prospecting, which is where the unworked audience lives. Most accounts that left the default uncapped were spending 40-55% on the same converters month after month.
The plateau is not a death certificate for Advantage+ Shopping. It is the end of the easy compounding phase. Accounts that treat creative velocity, catalog segmentation, and existing-customer budget caps as Q2 operational priorities will resume the curve. Accounts that treat the plateau as "ASC stopped working" will leave Q2 looking for a replacement that does not exist.
The Advantage+ vs Detailed Targeting decision tree still holds for non-shopping campaigns. Shopping is its own animal in 2026, and the moves above are what restart the engine. Server-side tracking through the CAPI auditor is the prerequisite either way.
Written by Foreground Digital. Start a project →