CONTENTS

    The Trade Desk’s Audience Unlimited Review (2025): Can AI-scored data actually cut your “data tax”?

    avatar
    Tony Yan
    ·September 30, 2025
    ·7 min read
    AI-scored
    Image Source: statics.mylandingpages.co

    If you’ve been frustrated by opaque third‑party data fees and mixed results from prebuilt segments, The Trade Desk’s new Audience Unlimited aims to change the calculus. According to the official Business Wire announcement, The Trade Desk announces a major overhaul of its digital advertising data marketplace (Sept 29, 2025), introducing AI‑scored audience selection and clearer pricing tied to trading modes. Disclosure: We are not affiliated with or compensated by The Trade Desk; this is an independent review.

    What is Audience Unlimited?

    In short, Audience Unlimited (AU) is an AI‑powered layer inside The Trade Desk’s Kokai platform that scores and ranks thousands of third‑party datasets from hundreds of providers, aiming to surface the most relevant audiences for your objective. The announcement describes support across channels including CTV, retail media, web, and mobile, along with a rollout beginning with select agencies in late 2025 and broader access in early 2026, as detailed in the The Trade Desk investor news detail on the launch (2025).

    Two AI “Koa Adaptive Trading Modes” sit alongside AU:

    • Performance Mode: Koa acts as a co‑pilot to optimize bids and allocation.
    • Control Mode: Buyers maintain more manual control over levers.

    Trade press coverage consistently echoes these changes and timing, with summaries such as the Campaign Asia breakdown of AI scoring and pricing tiers (2025) and the PPC Land recap of the data marketplace overhaul (2025).

    Pricing and modes: the unusually explicit part

    One standout element here is transparent pricing in Control Mode:

    Why this matters: Most DSPs bundle or negotiate third‑party data fees within curated packages, making apples‑to‑apples cost comparisons difficult. AU’s explicit percentages let you forecast the CPM impact with simple math (see below).

    Channels and availability

    Quick cost math: translating 3.3%/4.4% into CPM impact

    To sanity‑check the cost implications, convert percentage fees to a CPM add‑on:

    • At a $10 media CPM, a 3.3% AU fee adds $0.33 CPM; a 4.4% fee adds $0.44 CPM.
    • At a $20 media CPM, 3.3% adds $0.66; 4.4% adds $0.88.

    Historically, many a la carte third‑party datasets effectively added dollars to CPMs rather than cents. AU’s structure could reduce the “data tax” materially—but the real question is whether AI‑scored audiences also deliver equal or better performance (CPA/ROAS, reach quality) for your specific goals. As of late 2025, no public case studies quantify AU’s lift; evidence is largely limited to official descriptions and trade coverage (insufficient public performance data to cite).

    Identity and privacy context (UID2/EUID)

    Audience Unlimited lives inside The Trade Desk’s broader identity‑enabled ecosystem. While AU’s press materials don’t spell out identity specifics, it’s relevant context that The Trade Desk pioneered Unified ID 2.0 (UID2), an open, consented identifier framework designed for the open internet. For background, see the UID2 overview documentation (maintained by the UID2 community). This helps frame how deterministic identity can power addressable reach in channels like web, apps, and CTV. Note: the launch materials do not explicitly claim “AU integrates UID2”; treat UID2 as ecosystem context rather than an AU feature promise.

    How to test Audience Unlimited without guesswork

    You don’t need to wait for third‑party studies to evaluate AU. Structure a clean pilot and measure it rigorously:

    1. Design your arms per objective and channel
    • Arm A (Baseline): Your legacy third‑party segments or contextual only.
    • Arm B (AU): Enable Audience Unlimited targeting.
    • Run both in Performance Mode (AU included) and Control Mode (AU fee line item) to compare apples to apples.
    1. Normalize for fairness
    • Match budgets, creatives, flight dates, and geos; pause other experiments.
    • Keep frequency caps and pacing consistent.
    1. Measure what matters
    • Efficiency: CPM/eCPM, CPA/CAC, ROAS.
    • Reach quality: match rates or logged‑in signal coverage where available; overlap and deduplication.
    • Incrementality: use geo or time‑split lift methods when feasible. For a primer, see this explainer on causal lift using geo/PSA frameworks.
    1. Run long enough
    • 2–4 weeks minimum per channel with steady budgets to smooth auctions and delivery.
    1. Document everything
    • Record bid strategies, optimization goals, segment selections, exclusions, and brand‑safety settings to keep the test reproducible.

    Tip: Before you pick audiences, align targeting with your buyer personas and funnel goals. If you need a refresher on upstream audience thinking, this primer on SEO explained for audience intent offers helpful groundwork that applies beyond organic. For broader planning, see how to structure a CMI‑style marketing strategy to keep paid and owned efforts in synch.

    Strengths and trade‑offs

    What impressed us

    What gives us pause

    • Evidence gap: No public, third‑party‑verified case studies yet quantifying CPA/ROAS lift or reach quality improvements from AU.
    • Rollout timing: If you’re not among select agencies in late 2025, you may need to wait until early 2026 to fully pilot, as noted on the 2025 The Trade Desk investor news page describing the rollout timeline.
    • Mode dependence: Inclusion in Performance Mode is attractive, but buyers who prefer granular control will need to consider the Control Mode fee and ensure workflow fit.

    Comparison: Audience Unlimited vs DV360, Amazon DSP, Yahoo DSP

    Below is a practical, non‑exhaustive comparison using criteria that matter to media buyers. Because competitors’ fee schedules are often confidential or bundled, the emphasis is on transparency and workflow implications rather than exact price parity.

    CriterionAudience Unlimited (TTD)Google DV360Amazon DSPYahoo DSP
    Data pricing transparencyExplicit 3.3%/4.4% in Control; included in Performance ModeGenerally bundled/opaque within curated packages; partner add‑ons varyOften bundled or negotiated; public rate cards uncommonOften bundled/negotiated; limited public detail
    AI audience assistanceAU scores/ranks third‑party datasets; Koa optimization modesML‑driven features like Optimized Targeting and audience expansion (details vary by tactic)Amazon first‑party insights with predictive audiences; specifics varyIdentity and audience products exist; AI features vary
    ChannelsCTV, retail media, web, mobile (per launch)Display, video, CTV, audio; strong YouTube/partnersDisplay, video, CTV across Amazon‑owned and third‑party inventoryDisplay, video, mobile, CTV
    Identity contextLives within TTD’s UID2/EUID ecosystem context (not an AU‑specific claim)Google privacy‑safe signals; limited cookie relianceRobust logged‑in shopper identityYahoo first‑party identity signals
    Measurement transparencyStandard DSP reporting; AU‑specific lift evidence not yet publicMature reporting; partner integrationsMature reporting; retail media attribution optionsMature reporting; partner integrations

    Interpretation

    • Transparency edge: AU’s stated percentage fees are unusually explicit compared to rivals’ bundled approaches.
    • Performance unknowns: Without public benchmarks, any performance advantage of AI scoring over legacy segments remains to be proven in your own tests.
    • Workflow fit: If you embrace AI‑assisted optimization, Performance Mode’s “included” pricing simplifies evaluation. Manual‑leaning teams must assess Control Mode costs vs expected lift.

    Scoring rubric (evidence‑bound)

    Weights reflect what matters most to performance buyers. Where public evidence is missing, we note “Insufficient data.” These are provisional and should be updated after pilots.

    • Performance & Targeting Precision (30): Insufficient data (no public CPA/ROAS/lift studies yet). Provisional: 22/30 based on feature promise and ecosystem maturity.
    • Cost Efficiency (20): Clear fee structure is a plus; inclusion in Performance Mode is attractive. Provisional: 16/20.
    • Usability & Workflow Integration (15): Koa modes promise flexibility; real‑world learning curve unknown. Provisional: 11/15.
    • Data Quality & Scale (15): Broad provider curation is claimed; provider list and depth not public. Provisional: 11/15.
    • Privacy & Identity (10): Strong platform context with UID2/EUID, but not an AU‑specific feature claim. Provisional: 7/10.
    • Measurement & Transparency (5): Standard DSP reporting expected; AU‑specific lift transparency TBD. Provisional: 3/5.
    • Support & Availability (5): Staged rollout into early 2026. Provisional: 3/5.

    Provisional total: 73/100 (subject to change with real‑world results). Note: Scores will be revised when independently auditable data emerges.

    Who should pilot Audience Unlimited now?

    Recommended to pilot if any of the following apply:

    • You’re paying meaningful third‑party data fees and want a more predictable, transparent structure.
    • You run cross‑channel buys (CTV + retail media + web) and want to centralize audience selection.
    • Your team is open to evaluating Performance Mode (AI co‑pilot) vs Control Mode side by side.

    Consider waiting or limiting scope if:

    • You require public case studies before testing new workflows.
    • Your governance requires full manual control and you’re sensitive to any percentage‑based add‑ons.
    • Your 2025/early‑2026 campaign calendar can’t accommodate a 2–4 week controlled pilot.

    Bottom line

    Audience Unlimited is one of the clearest attempts to demystify third‑party data buying in a major DSP. The explicit 3.3%/4.4% Control Mode fees—and inclusion in Performance Mode—make cost modeling refreshingly straightforward. The open question is performance: will AI‑scored segments consistently beat your baseline audiences on CPA/ROAS and reach quality? With no public benchmarks as of late 2025, the prudent move is to run a structured pilot in both Koa modes, track cost‑per‑outcome rigorously, and keep your procurement path flexible until you see lift.

    If AU’s AI ranking materially improves efficiency at cents‑on‑the‑dollar CPM add‑ons, it could become a default switch‑on for many campaigns. If not, you’ll still have learned precisely where AI‑assisted audience discovery fits in your playbook—without guesswork.


    References and evidence anchors in this review include the 2025 Business Wire launch press release detailing AU and pricing, The Trade Desk’s investor news page outlining modes and rollout, Campaign Asia’s coverage of AI scoring and pricing tiers, PPC Land’s recap of the overhaul and timing, BestMediaInfo’s summary of pricing and modes, and the UID2 community’s overview documentation for identity context.

    Loved This Read?

    Write humanized blogs to drive 10x organic traffic with AI Blog Writer