Spark & Affiliate Commerce Flywheel

Why it works right now

Campaign architecture

• Campaign 1 — Spark prospecting (50–60%): website-conversion objective, broad targeting, 8–12 live Spark creatives from brand handle + creator authorizations; daily budget ~10x target CPA per ad group^T37,T38.
• Campaign 2 — GMV Max (20–30%, if on TikTok Shop): mandatory buying mode for Shop^T19; judge on incremental GMV and cost per order, not its blended ROAS^T20.
• Campaign 3 — Search Ads (5–10%): keyword campaigns on category/product terms — 2–3x feed CVR at $1.50–3.00 CPCs^T24.
• Affiliate program (parallel, not media budget): targeted creator outreach with samples + commission; open collabs churn ~58% of creators in 90 days — managed programs outperform^T30.
• Smart+ test cell (optional): one Smart+ vs a manual control; watch for budget-burn and auto-edited creative — its −36% CPA claim has no independent validation^T16,T34.

Creative & offer requirements

• Floor: 8–12 live variants per campaign + 3–5 new versions per ad group weekly^T36,T4. Refresh weekly at $200–1K/day; every 3–5 days above $1K/day^T35.
• UGC-style native video only; polished studio assets are structurally disadvantaged (~28% worse CPA reported)^T4. Modular production: swap hooks/CTAs, don't reshoot.
• Refresh triggers: 3-second view rate <12% or CTR −30% week-over-week^T36.
• Offer: impulse price point or strong intro offer; >$200 AOV items are ~9% of Shop sales — run a funnel or don't run TikTok^T4.

Setup checklist

• Pixel + Events API; catalog synced; Shop linked if applicable; unified Shop+website conversion reporting on^T26.
• Spark authorization workflow with creators (codes collected, 30–60 day windows).
• Affiliate program live in Shop with commission tiers; sample logistics owned by the client.
• Measurement: incrementality or MMM check for budget decisions — in-platform ROAS is the least trusted read per 2026 industry surveys^T26,T27.

Expected performance ranges (benchmarks, not guarantees)

Timeline

1. 01

   Week 1–2: setup, creator authorizations, launch Spark prospecting. Usable signal in 7–10 days^T5.

2. 02

   Week 3–4: first refresh cycles; GMV Max live; kill weak hooks.

3. 03

   Month 2: stable performance window begins (30–45 days)^T5; Search Ads + Smart+ test cells.

4. 04

   Month 3+: affiliate flywheel contributing; judge on blended CAC + incremental GMV. Success gate: CPA at/below Meta equivalent for the same product with creative machine sustainable.

Scaling & optimization playbook

• When it works: scale creative supply first (AI variants via Symphony, more affiliates), then budget in ≤30% steps; push winning organic affiliate posts into Spark immediately.
• When it doesn't: check creative fatigue metrics before anything else (it's the #1 constraint — decay in 48–72h at high spend)^T35; then landing/Shop conversion path; then price point fit.
• Key levers: creative velocity, Spark vs dark mix, affiliate commission tiers, catalog/price architecture.

Risks & failure modes

• 1 · The creative machine stalls. Creative starvation kills TikTok accounts faster than budget starvation^T4. Prevent: lock a weekly production commitment (client, creators, or AI-assisted) in the scope.

• 2 · Wrong product economics. High AOV, older audience, considered purchase^T4,T6. Prevent: qualify with the wrong-fit list before proposing; below ~$2K/month total paid social, fund Meta first^T40.

• 3 · GMV Max self-flattery. Blended organic+paid attribution inflates ROAS^T20. Prevent: baseline organic GMV pre-launch; report incremental GMV.

• 4 · Algorithm drift under the JV. Performance shifts as the US model retrains^T12. Prevent: re-run incrementality after major swings; keep Meta as the stable comparator.

Minimum viable version