Meta Advantage+ and Google Performance Max: how AI automates campaigns and what advertisers still control

Meta's Advantage+ (formerly Advantage+ shopping, now Advantage+ sales) and Google's Performance Max are campaigns where the platform's AI decides who sees the ad, in which placement, with which creative combination and at what bid. The advertiser sets the goal, supplies creatives and data, and sets budget. It's the deepest shift in digital advertising this decade: manual targeting loses weight, while data and creative quality gain it.

What each one automates

• Targeting: instead of closed audiences, AI looks for buyers across the inventory using signals.
• Bidding: real-time adjustment based on each impression's conversion probability.
• Placements: Facebook, Instagram, Reels and Stories on Meta; Search, YouTube, Display, Discover, Gmail and Maps on Google.
• Creatives: combines headlines, copy, images and videos and prioritizes the best-performing combinations.

What the platforms say

Google states in its help center that advertisers adopting Performance Max see on average 27% more conversions or value at a similar CPA or ROAS, even if they already used broad match and Smart Bidding on Search. In 2025 it also cited 18% more valuable conversions versus traditional campaigns. Meta presents Advantage+ as an end-to-end solution optimizing creative, targeting, placement, budget and destination.

What independent studies say

Haus, a company specializing in incrementality measurement, analyzed 640 Meta experiments over 18 months. It concluded Meta is incremental in most cases (19% average lift on the primary KPI). On Advantage+, the picture is more nuanced: on average it did 9% better than manual campaigns at the experiment midpoint but 12% worse by the end. Its reading: Advantage+ is very good at capturing existing intent, but should be balanced with mid- and upper-funnel strategies.

What the advertiser still controls

• The goal and conversion value: optimize for a weak event and AI optimizes badly.
• Data quality: Meta's Conversions API, Google's enhanced conversions and real order values.
• Creatives: the main lever; AI can only choose among what you give it.
• Exclusions: brand terms, existing customers or placements you don't want.
• Budget and structure: Google recommends fewer, larger campaigns with at least 15 conversions a month for target ROAS.

If creative is the lever, variant volume matters. AI video tools and prompts for ad copy help produce enough versions for the algorithm to choose from.

How to test it in your account

1. Fix measurement first: correct conversion events with value.
2. Split budget between automated and manual, with a control group.
3. Let the test run several weeks without abrupt changes.
4. Evaluate with incrementality (geo or holdout test), not just platform ROAS.