Algorithmic Audience Targeting in Meta Ads Remarketing Campaigns: A Demographic Analysis of AI-Driven Delivery in a Romanian Aesthetic Medicine Clinic

Within the Meta advertising ecosystem (Facebook and Instagram), audience targeting decisions are increasingly delegated to machine-learning systems that determine, in real time, which users see an ad and at what frequency. This paper analyzes the demographic distribution of algorithmically delivered impressions in a remarketing campaign run by a Romanian private aesthetic medicine clinic, a real client from my digital marketing agency (Performance Target) over a 122-day observation window. Using ad-set-level data exported from Meta Ads Manager, with breakdowns by age and gender, the study examines how a single remarketing campaign - nominally addressed to the entire pool of website and profile visitors - is in practice distributed across demographic segments by the platform's delivery algorithm. Results show a very strong gender skew (94.9% of the budget delivered to women) and a clear age concentration: women aged 35 to 54 receive 58.3% of total spend, and women aged 35 to 64 receive 76.5%. The cost per thousand impressions (CPM) ranges between 13.18 and 17.70 RON depending on segment, while the campaign maintains a low average frequency of 1.10 to 1.14 over the entire window, indicating that the algorithm continuously identifies new audience members rather than over-exposing the same users. The findings illustrate, in a healthcare service setting, the dynamics of algorithmic ad delivery documented in the broader literature on AI-driven advertising, with implications for managerial control over targeting, for advertiser accountability and for the design of marketing strategies in regulated service sectors.