trendbuzz.to
← All posts
Targeting · Dec 18, 2025 · 10 min

Cookieless Targeting in 2026: What Actually Works

Imani Okafor · Head of Data Science

A field report from 240 campaigns across CPG, fintech, gaming and travel. Contextual + on-device cohorts deliver 92% of the lift of legacy 3P identity.

Programmatic display, native, video and Connected TV continue to consolidate around a smaller set of high-quality supply paths. In 2026, three trends matter more than the others: curated PMPs as a default buying surface, contextual signals overtaking 3P cookie identity, and the rapid normalization of CTV ad-pod auctions.

1. Curated PMPs become the default

Curated marketplaces moved 41% of net open-exchange spend in Q4 2025, according to TrendBuzz internal data, up from 27% in the same quarter a year earlier. Buyers are increasingly using curated packages as their primary buying surface rather than as an experimental layer.

The decision driver is simple: bid-request quality. A curated package surfaces only the supply that meets brand suitability, viewability, contextual and quality criteria pre-bid. Buyers see fewer bid requests but win a higher percentage of those they bid on, leading to a typical 18–24% improvement in working media.

2. Cookieless targeting catches up

Field data across 240 campaigns shows that a contextual + on-device cohort blend now delivers 92% of the lift produced by legacy third-party identity stacks. For CPG and travel verticals the gap closes to 96%. The remaining gap is in long-tail retargeting use cases, where probabilistic methods still trail deterministic IDs.

3. CTV ad-pod auctions normalize

Ad-pod CTV auctions — where a single VAST pod is filled by multiple winners across competitive separation rules — are now the standard for SSAI-stitched inventory. The change forces buyers to think about pod position, not just CPM, and forces SSPs to expose deduplicated pod outcomes.

What to do about it

  • Move always-on lines to curated PMPs first; reserve open exchange for prospecting and tail reach.
  • Migrate identity-heavy audiences to contextual + cohort, and run a parallel reads test against your MMP for at least 30 days.
  • For CTV, configure pod position and competitive separation rules explicitly in your DSP, and require log-level pod outcome reporting from your SSP.

Have a different field observation? Send it to [email protected]. We publish guest analysis once a month.