Toku Health · creative formats · Oct 2026

Everything we can make for Toku

A menu of ad formats for our brainstorm. Each card is either an ad I edited for another DTC brand (play it) or a new format worth testing, with the data behind it. Under each one is the Toku version and what it takes to make: AI only, a re-cut of footage we already have, or a creator shoot.

01 · Format library

Formats, grouped the way we'd test them

A format is the container; the angle (ED, Stress, Heart Health) is what we test inside it. Groups follow the standard DTC format taxonomy. Examples come from brands I've edited for (Everyday Dose, RYZE and others); spend tags are lifetime Meta spend on Everyday Dose.

Needs: Show:

02 · What's working

What the data says right now

Market benchmarks for 2025–2026. Use these to pick which formats get budget first, not as guarantees.

Hit rate by format

Share of creatives in a format that became winners (≥10× the account's median spend and ≥$500).

Motion Creative Benchmarks 2026 (vendor data): 578,750 creatives, 6,015 accounts, $1.29B Meta spend, Sep 2025–Jan 2026. Source

Health & wellness: top formats by hit rate

StitchReaction videoUnboxingCelebrityFounderLetterStop motionInfluencer endorsementPOV

In order, from Motion's health & wellness cut of the benchmark.

Motion, visual formats by vertical. Source

Winning supplement ads

  • 62% of winning videos run 30–60 seconds
  • 52% open with an authority hook
  • 67% of winning videos are founder-led UGC
  • Winning statics break down ingredients with doses

Brandsearch, Apr 2026 (vendor data; "winner" = running 30+ days). Source

−45%

Partnership ads

Median cost per incremental purchase in Meta's 20 lift tests (EMEA, Jan–Jun 2025). A separate study of 65,000 ads found +19% CTR and −5% CPA vs. licensed UGC.

PPC Land · Net Influencer / Agentio

42%

Lo-fi wins

Share of top-performing DTC ads that are deliberately low-production, while lo-fi is only 14% of all ads.

Motion 2025 via Net Influencer

61%

Statics still carry spend

The median account runs 61% static and 39% video (67,000 ads analysed). Statics are the cheapest way to test an angle before it earns video budget.

Curtis Howland via Segwise

03 · Test plan

How I'd run the first month

A proposal to agree on together. It fits our rhythm (briefs Friday, production Monday) and the 20-pieces-a-month base.

Angle first, format secondProve which angle sells with cheap formats, then spend on bigger executions of the winner.
AI first, creator secondTest a concept as AI VO or animation. If it works, brief a real creator on the same script.
1 concept = 1 ad set3 creatives per concept, each with 2 headlines and 2 primary texts. Run 7 days.
Kill fastTurn off at 2–3× target CPA with no purchase. Hook rate under 25% is a kill, over 40% is a scale signal.
Diversity over hook tweaksMeta now rewards different formats and faces. Near-identical AI hooks with the same face stop getting delivery.
80 / 20 once we have winners80% iterations on winners, 20% net-new concepts.
Week 1–2 · Find the angle

Cheap formats × 3 angles

  • AI VO over library B-roll: ED, Stress, Heart Health
  • Pixar story: one per angle
  • 6–10 bumpers across angles
  • Long-form static per angle (with the statics team)
Week 3 · Expand winners

Same script, new executions

  • Creator version of the winning script (2–3 creators)
  • Person swaps of the winner
  • Voiceover ↔ text-on-screen flip
  • AI song or animated ingredients version
Week 4 · Authority + offer

Make it hard to copy

  • Podcast-style shoot with Justin or an expert
  • Founder-led clip on why Toku exists
  • Offer banner on the top 3 ads
  • Partnership ads from creator handles
MetricWhere it mattersRule of thumb
Hook rate (3s views ÷ impressions)Top of funnel, first read on a new format<25% kill · 30–40% good · >40% scale
CTRTop and middle of funnelCompare within the same concept
CPA / ROASFinal call on every conceptA high hook rate doesn't pay bills; conversion decides
Share of spendCalling a real winnerSpend + KPI + time, not a 7-day spike

Previews are compressed to 480p. Market figures are vendor data and describe averages across accounts, not Toku. Every script goes through Justin's compliance review before production. — Ross