AI Audience Segmentation Adoption 2025
Tags: Statistics, Trends
TL;DR
- 51% of agencies and 35% of publishers use AI for audience segmentation; many also deploy synthetic data to offset signal loss.
- 25–33% are using generative AI to build segments with synthetic (fake) data.
Why it matters for HK marketers: AI‑driven segments boost addressability without third‑party IDs, but raise profiling and discrimination risks.
What’s happening
- Adoption: Agencies lead, publishers catching up; brands also investing.
- Why now: AI helps rebuild addressability post‑cookie by inferring intent from content and first‑party signals.
- Synthetic data: Used for modeling and testing when first‑party data is thin or sensitive (e.g., minors, health) — but watch for homogeneity and validation gaps.
51% of agencies use AI for segmentation; 35% of publishers do.
One‑quarter to one‑third use genAI to build segments with synthetic data.
Risk focus
- Profiling exposure: Inferences can reveal sensitive traits (e.g., health, religion, emotion), triggering regulatory scrutiny.
- Label hygiene: Segment names and logic need review to avoid sensitive or discriminatory proxies.
So what for marketers
Institute a segment review process, limit sensitive inferences, and document data sources/consents — especially when synthetic data informs targeting.
← Back to Knowledge Base