Olivia Kory, Chief Strategy Officer from Haus, walks through an 18-month review of 640 incrementality tests run by Haus to compare Meta’s Advantage Plus campaigns with manual setups. She explains the shift from hands-on targeting to AI-led buying, why incrementality matters more than platform-reported conversions, and how Geolift holdouts estimate true lift without user-level data. The results show a strong overall impact from Meta, but mixed performance from Advantage Plus versus manual, along with signs that mid-funnel optimization can change outcomes. She ends with practical cautions on interpreting automation results and the need to test by brand. Takeaways - Incrementality testing helps separate real lift from conversions that would have happened anyway. - Meta tends to deliver measurable lift quickly, though attribution settings change what you see. - Advantage Plus performs inconsistently against manual campaigns and can skew toward low funnel intent. - Mid-funnel signal engineering may broaden reach and improve omnichannel effects, but needs validation. 00:00 Overview of the research question on AI buying and outcomes. 03:10 How Meta buying moved from manual controls to automated setups. 06:05 Why incrementality matters more than reported conversions. 09:00 Geolift holdout method and cross-channel measurement. 12:10 What the wider Meta test set shows about lift and timing. 16:05 Advantage Plus versus manual results and efficiency gaps. 19:20 Possible reasons for the Advantage Plus pattern and open questions. 21:10 Signal engineering and mid-funnel optimization approach. 24:20 Final cautions and how to apply testing to your own brand.