Video Viewer Audiences: Building Warm Pools for Retargeting

Video retargeting is worth funding when the complete journey earns more retained customer contribution than a simpler direct campaign. Cheap plays alone cannot answer that question. Define the video cohort, keep repeated plays separate from unique eligible people, and compare the costs of every stage before expanding the funnel.

Define the denominator before pricing the audience

Record selected video IDs, runtime, accepted audience threshold, window, data source and report timestamp. The Meta CustomAudience SDK exposes video subtype, approximate count bounds and delivery status; it does not turn view totals into an exact reachable pool. Check the actual audience creation options and saved campaign controls. If only a range or unavailable count is returned, report it that way and leave exact member cost undefined.

Watch depth measures exposure, not qualification

Hypothetical runtimes: 75% of a 20-second clip equals 15 seconds, while 75% of a two-minute demo equals 90 seconds. Neither is a completed 100% view or a qualified lead. A short playback may contain the entire offer; a deep viewer may be researching, entertained or already a customer. Treat “deeper exposure may help this offer” as a hypothesis and verify qualified leads or retained payments downstream.

Illustrative arithmetic, not a market benchmark: $500 at $0.01–$0.05 per recorded play buys 50,000–10,000 plays respectively. Even 50,000 plays do not prove 50,000 unique people. If a separate verified unique-viewer report hypothetically returned 8,000 people, cost per unique viewer would be $500 / 8,000 = $0.0625; eligible audience size could still differ. Never combine a playback price with a promise of unique pool size.

A two-route brief tests whether warming is necessary

Hypothetical hypothesis: a tutorial helps customers understand whether the organizer fits their desk, enough to cover its extra cost. Use accurate measured dimensions and show the actual product. Proposed follow-up copy: “Check the organizer dimensions and see what fits before you order.” Do not fabricate urgency or customer testimony. Verify supported source exclusions; sequence instructions do not guarantee each individual sees the tutorial before the offer.

Approve one total allowance per route and a common follow-up period before testing. Use eligible randomized assignment where available; if delivery groups overlap or differ materially, call the comparison descriptive. Include direct purchases after the tutorial as well as retargeted purchases, but count a customer once. Document organic purchases, unresolved attribution and refunds. A holdout with a planned sample and uncertainty is needed to claim incremental impact, rather than comparing warm buyers with cold prospects.

Calculate the whole journey

Illustrative assumptions, not observed results: the warming route spends $200 on video distribution, $300 on retargeting and $100 on production, and gets 15 unique retained new customers. Full acquisition cost = $600 / 15 = $40. A direct route spends $500 on media and $100 on production and gets 18, for $600 / 18 = $33.33. If each retained customer contributes $35 before acquisition cost, warming contribution after acquisition is 15 × $35 − $600 = −$75; direct is 18 × $35 − $600 = $30. The hypothetical cheaper-view route therefore loses on this business outcome.

• Choose reporting age from actual purchase and refund delay; do not compare a one-day warm cohort with a month-old direct cohort.
• Review at the approved total allowance or date, and stop sooner for invalid permissions, broken targeting or an unsafe offer. Pending outcomes justify recording uncertainty, not automatic extra spend.
• If the direct route is sufficient, skip the extra warming stage. If both fail, investigate offer, fit, checkout or measurement rather than purchasing more views.

Prepare and review the campaign

Use the two-route brief to prepare a NovaStorm campaign draft and inspect its saved objective, audience, budget and creative before publication. Keep a separate defined-metric worksheet with media, production, unique retained payments and reporting age; do not assume automatic blended-funnel reporting. If the remaining question is watch-depth selection, use engagement audience logic to check unions, overlap and runtime first.