The format, as measured in the data
Our Apify pull of Atolea's 300 active ads shows a stark format split that directly contradicts agency defaults:
| Format | Atolea active ads (%) | Typical DTC brand (%) | Why the gap |
|---|---|---|---|
| Video | 57% | ~40% | Meta's algorithm actively prefers video since 2024; Atolea over-indexes deliberately |
| Image (static) | 43% | ~35% | Single-image, high-contrast — used for retargeting and broad-test stages |
| Carousel | 0% | ~20–25% | Atolea deliberately avoids carousels — they don't convert in DR use cases |
Source: ad-intelligence/brands/atolea-jewelry.json; FOTW video breakdown with Adil Mourjane (funnels/atolea-jewelry/video-transcript.txt).
The 30-second reaction-video anatomy
Adil's breakdown of the winning Atolea video ad (from the FOTW video-transcript) dissects the structure second-by-second. Every piece is load-bearing:
| Seconds | On screen | Purpose |
|---|---|---|
| 0–2 | Tight b-roll: jewelry being put on / worn during activity | Pattern interrupt. Hand + jewelry + motion = stop-scroll |
| 2–4 | Caption: “I was worried about…” | Pain-point articulation — the target demographic's internal hesitation |
| 4–8 | Caption continues: “…tarnish / green skin / losing it in the shower” | Specific, named fear. No generic “cheap jewelry problems.” |
| 8–14 | B-roll: shower / pool / workout / sleeping in jewelry. No voiceover. Natural sound only. | Demonstration of mechanism — showing what the customer thought was impossible |
| 14–20 | Caption: “I've worn this for 8 months. No tarnish. No problems.” | Payoff — the mechanism worked |
| 20–26 | B-roll: same jewelry, later — outside, sunlit, still looking new | Proof of durability over time |
| 26–30 | CTA card: product + “Shop Now” + Lifetime Color Warranty badge | Close with the trust anchor that neutralizes residual hesitation |
“Zero voiceover is deliberate. The moment you add voiceover, it becomes an ad. Without voiceover, it's just someone showing you something they own. The captions do the selling, the b-roll does the proof, and the customer does the thinking.”
— Adil Mourjane, FOTW video breakdown of Atolea (transcript at ~12:40, funnels/atolea-jewelry/video-transcript.txt)
Direct translation to 96NORTH candles
Atolea's pain-point captions work because they name a specific anxiety the customer hasn't admitted to themselves. For candles, the equivalent anxieties are:
- “I was worried it would give me a headache.” (Paraffin / synthetic fragrance anxiety)
- “I was worried it would smell artificial.” (Cheap fragrance oil fear)
- “I was worried it would tunnel and burn out in 10 hours.” (Wax quality fear)
- “I was worried the scent wouldn't fill the room.” (Underperformance fear)
- “I was worried it would look cheap on my shelf.” (Aesthetic / giftability fear)
Each of those becomes the opening caption of a 30-second reaction video.
The 15-creative production slate
Atolea's content library sizing: ~60 clips shot across ~8 creators + in-house produces ~300 active ads through remixing (headline swaps, length cuts, B-roll rotation). The inverse math: to produce 15 ad variants, you need ~3 clip-days of production, ~10 clips per day, ~30 clips total.
| # | Angle | Opening caption | B-roll spec | Payoff caption |
|---|---|---|---|---|
| 1 | Headache fear | “I was worried this would give me a headache.” | Candle burning on nightstand, person reading, tranquil | “It's 100% natural soy. No synthetic fragrance. I burn it every night.” |
| 2 | Artificial scent fear | “I was worried it would smell like a department store.” | Close-up of wax pool + oils; then room context | “Aromatherapy-grade French lavender. It smells like the actual field.” |
| 3 | Tunnel fear | “I thought it would tunnel after 10 hours.” | Candle across 3 days of use, full melt pool edge-to-edge | “50 hours later. Still edge-to-edge. Still scented.” |
| 4 | Room fill fear | “I was worried it wouldn't fill the room.” | Wide shot of room, candle small in frame, then close-up of person inhaling | “3-wick. Aromatherapy oils. The whole apartment, not just one corner.” |
| 5 | Giftability fear | “I was worried she'd think it was a cheap gift.” | Unboxing of the gift box, person's reaction, candle on coffee table | “She texted me the next day asking where I got it.” |
| 6 | Price objection | “$26.95 sounded expensive for a candle.” | Side-by-side: 96NORTH vs. $8 drugstore, both burning at hour 10 | “Tunneled in 8 hours. Still burning clean at 50.” |
| 7 | Paraffin fear | “I didn't know what paraffin wax actually was. Until I did.” | Text overlay on ingredients label; pan to 96NORTH “100% soy” | “100% natural soy. No petroleum. Burns clean.” |
| 8 | Routine integration | “My evening before… my evening now.” | Split: chaotic / with candle lit, slower pace | “Five minutes to light. The whole apartment settles.” |
| 9 | Quality doubt | “I was skeptical of ‘aromatherapy’ candles.” | Macro shot: raw lavender oil, raw soy wax, wick being set | “These are the ingredients. Not fragrance. Actual oils.” |
| 10 | Amazon comparison | “Amazon has 10 candles under $15. Here's why I still pay $26.95.” | Multi-candle lineup; pan across; focus on 96NORTH | “50-hour burn. 100% soy. No headaches. Made to last.” |
| 11 | Burn-time proof | “This candle has been burning on my shelf every evening for 8 weeks.” | Time-lapse: candle across multiple evenings, same place | “Still 2/3 full. $26.95 → $0.53/night.” |
| 12 | Gift recipient POV | “The best gift I got this year was $26.” | First-person unboxing; smelling the candle; lighting it | “96NORTH French Lavender. I've asked for two more.” |
| 13 | Seasonal hook (Q4) | “My Halloween is a pumpkin spice candle and no one in my house.” | Cozy sofa, candle, book, blanket | “50 hours of pumpkin spice. October has a smell.” |
| 14 | Seasonal hook (Dec) | “Christmas doesn't start until I light this.” | Tree, cozy room, candle foreground | “Winter Pine. One candle. The whole house.” |
| 15 | Sub-teaser | “2 candles every month, $35.90, shipped on the 1st.” | Unboxing member box; two Pure candles on table | “Skip any month. Cancel with two clicks. (waitlist open.)” |
#15 is the Idea 01 teaser — bridges into the membership waitlist once the subscription is live.
Format rules (non-negotiable)
- Zero voiceover. Captions only. Music = simple, low-BPM, no vocals.
- Vertical 9:16 master, with 4:5 and 1:1 cut-downs. Meta placements: Reels + Stories are 80% of the delivery; feed takes the 1:1.
- Caption on-screen at all times, highly legible. Over 80% of Meta video views are sound-off.
- First frame is the hook. If the first frame isn't a pattern-interrupt, redo the shot. No brand logos in the first 2 seconds.
- No CTA overlay until second 26. Captions earn the CTA. Premature CTAs kill watch-through.
- Same b-roll, many caption sets. Each 8-clip production day yields 40+ final variants after caption remixing. This is how Atolea runs 300 ads on 60 source clips.
The Goose Creek corroboration
Second data point: Per the Manus AI PDF and our own Apify pull, Goose Creek runs the same format structure at ~$400K/quarter Meta spend with 3.0+ ROAS. Goose Creek is the closest direct competitor to 96NORTH by price point ($15–$28 candles, US DTC) and their creative playbook is almost identical to Atolea's: no-voiceover b-roll + caption overlays + occasional UGC. Two independent brands converging on the same format is the strongest signal we can give that this isn't an Atolea fluke.
Production cost + timeline
| Item | Cost | Time |
|---|---|---|
| Creative director / briefs (in-house) | $0 (or $2K freelance) | 1 week |
| 2-day shoot: b-roll only, 5 locations (home bedroom, home bath, home office, outdoor, kitchen) | $3–5K (DP + prop + travel) | 1 week |
| Editor: 15 final variants from ~60 source clips + caption sets | $2–3K | 1–2 weeks |
| Music licensing (royalty-free pool: Artlist / Epidemic) | $200 | — |
| Total | $5–10K | 3–4 weeks |
Success metrics
- Hook-rate: 3-second-view-through / impressions. Target > 30% (Atolea benchmark).
- Hold-rate: 15-second-view / 3-second-view. Target > 50%.
- CTR: target 1.5–2.5% cold traffic, 3–5% warm.
- Cost per thumbstop: $0.02–0.04 on Reels placements if format is tight.
Evidence gaps
What we can show vs. what we can't. We have the structural data (57/43/0 split, active ad count per page, format distribution) and the FOTW video transcript with Adil's second-by-second breakdown. What we don't have is the rendered ad creative in our R2 library — Atolea's specific video files aren't in our media-index because they're hosted on Meta's servers and change weekly. If you want to see examples before producing, pull them live from Meta Ad Library → Atolea — the format pattern is visible across almost every active video ad.