YouTube Halo Engine

Why it works right now

Campaign architecture

• Demand Gen — Create (50–60% of program): multi-format (video + image), Google O&O channels only, vertical assets included for Shorts delivery inside multi-format (never Shorts-isolated: 20% lower iROAS at 32% higher CPMs)^Y9,Y4. Maximize Conversions → tCPA at 30+ conv/month^Y20.
• Capture net (30–40%): branded search + Shopping/PMax sized up ~15–25% for the demand wave; Amazon presence stocked and optimized (that's where the halo lands)^Y4,Y18.
• Measurement (5–10% of program cost): geo-lift/holdout design from launch; branded-search conversion-rate tracking^Y19; readouts on 4–8 week windows (lift kept building +79% after campaigns ended)^Y4.
• CTV: included as an additive surface inside Demand Gen at same ROI target — never budget-forced for DR (65% lower iROAS when forced)^Y4.

Creative & offer requirements

• 3–5 video concepts per quarter × 15s/30s cuts × horizontal + vertical; hooks in the first 3 seconds for CPM surfaces; price/eligibility pre-qualification on CPC surfaces^Y9.
• Image assets alongside video (Google claims +20% conversions for multi-asset; treat as directional)^Y7.
• Creator-style and founder-story content outperforms polished spots for DTC; Creator Partnerships can license creator videos as partnership ads^Y14,Y15.

Setup checklist

• Enhanced conversions, consent verified; new-customer conversion goal configured.
• Channel controls set to Google O&O; account-level placement exclusions^Y9,Y11.
• Geo-lift test design (matched markets, 4–8 week windows) agreed with the client pre-launch; Amazon sales data access for halo reads^Y18.
• Brand search + Shopping capture campaigns verified healthy before video spends a dollar^Y9.

Expected performance ranges (benchmarks, not guarantees)

Timeline

1. 01

   Week 1–2: capture-net hardening, creative finishing, geo-design; launch Demand Gen.

2. 02

   Week 3–4: delivery QA (placements, view quality); no performance judgment yet.

3. 03

   Month 2: first lift readout window opens (4–8 weeks); branded-search and Amazon deltas reviewed.

4. 04

   Month 3+: scale/hold decision on iROAS; quarterly testing cadence institutionalized^Y5. Judge at month 3, not before.

Scaling & optimization playbook

• When it works: scale spend 20–30% per flight while re-testing lift; expand concepts, geos, and Creator Partnerships; add Attributed Branded Searches as a directional dashboard metric^Y8.
• When it doesn't: check capture-net leakage first (out-of-stock Amazon, weak brand SERP), then creative hooks (view rate <20%), then audience signals. Kill only after a clean 8-week window.
• Key levers: creative hooks, O&O channel purity, capture-net readiness, test cadence.

Risks & failure modes

• 1 · Judged on last-click. YouTube always loses that report (click CVR 0.05–0.5%)^Y3. Prevent: incrementality framing in the contract; decline the engagement if the client insists on CPA parity with brand search^Y4.

• 2 · No capture infrastructure. Demand created, nowhere to land^Y9. Prevent: S6/S9-grade foundations first; Amazon stocked.

• 3 · Underfunded test. Below ~$3K/month video, significance never arrives^Y21,Y9. Prevent: gate the proposal on budget.

• 4 · Creative that can't hold 3 seconds. View rate collapses, CPMs buy nothing. Prevent: hook testing in cheap in-feed placements before scaling in-stream.

Minimum viable version