## 1. Executive Summary
Jairet Crum is Performance Manager at TNT Media (Tried and True Media), a direct-response agency with nearly 80 staff that scales some of the largest offers in the world and only takes clients launching in the US market. An English teacher in Florianopolis, he discovered copywriting while doing startup marketing, freelanced for eight months, and joined TNT as its second copywriter in February 2019; five years later he built and ran the creative production system, was promoted to Creative Director, and now owns performance for a portfolio of clients. This episode is a field guide to his process: where elite US copywriters actually learn, how a systematic creative-production line works, the exact variation tests (hooks, sounds, formats, camera depth, settings) used to turn one validated ad into a weekly ad spend that grew from $10k to $200k in five weeks, how he validates offers against funnel benchmarks, and his highest-leverage VSL experiments cut-first editing, mobile-first formats, and stripping the sales page down to just the video.

## 2. Alpha / Sauce
- **Ninety percent of the job is ads, and the goal is to spend.** Jairet's performance role is accountable to one metric: making the ads scale grow volume at the client's target return on ad spend, then move to the funnel only when a metric is broken.
- **Respect the conversion.** When a method works, stop radical testing and squeeze it: TNT scales so aggressively they call it "squeezing the lemon" a single winning ad went from $10k to $200k a week in five weeks because variations rolled out before fatigue killed it.
- **One in five variations wins (80/20).** If 35 tests fail and seven work, the seven pay back tenfold stop mourning the losers and keep feeding the winners.
- **Elite copy comes from coaches and mastermind groups, not courses.** Their writers study inside group feedback loops (Copy Accelerator, Nothing Held Back, Fast Forward) and are mentored weekly by a classic-era copy coach plus the 9-sales Makepeace framework as an outline.
- **The first three seconds ARE the ad.** Test 3-second clips, era-appropriate sound effects (a record-player hiss lifted an ED offer; a fart sound lifted digestion), and a banner overlay all before touching a single word of the body copy.
- **A human face multiplies conversion.** A person speaking on camera, or a split-screen of eyes next to the eye-catching clip, exploits the instinct to look at a gaze; same message, same copy the face wins.
- **Cut first, never rewrite first.** The rare VSL that lost viewers at the 8-minute mark was fixed by cutting that section outright a 15% gain that, at high volume, is enormous.
- **80% of traffic is mobile.** A 16:9 desktop VSL loses people; re-render to 4:5 blurred-pillar, then full-bleed 4:5, then full-screen 9:16 each step beats the horizontal original.

## 3. Methods & Setup
1. **Build the creative production line instead of hand-writing ads.** Jairet wrote scripts as a two-column doc (copy on the left, every image choice on the right) so the editor only executes faster to a final product and easy to remix later. Setup: swipe files of running competitor ads, transcribe, and reuse proven elements.
2. **Systematize with a tracking tool and deadlines.** Jairet built the department around Trello (later ClickUp) with copy due-dates for review, then hand-off to a video editor or to stock-image/actor production. Setup: define pipeline stages: copy review edit publish measure.
3. **Follow the Makepeace method as a guide, not a template.** A VSL must make nine sales: attention, interest, credibility, opportunity/problem (root cause), solution, value, security, convenience, scarcity; RMBC (the Stefan Georgi outline) provides the tools to hit each objective. Setup: score every script section against its objective before and after testing.
4. **Measure 3-second engagement instead of guessing hooks.** TNT now judges the first three seconds by engagement rate and tests hooks as the variable "tired of tight clothes? scared of the scale? afraid to drive at night?" with the same body copy. Setup: run new hooks in batches of five.
5. **Run structured variation tests on a winning ad.** Immediately produce: four to five clips in the first 3 seconds, five sound variants, five music variants, a banner/headline overlay, then five to ten new hooks from a copywriter. Setup: test in groups of five for clean attribution.
6. **Test era-appropriate audio for aging buyers.** For an ED offer whose buyers are 60-70, a plain record-player or radio static sound in the first three seconds lifted sales no copy at all, nostalgia matching their "best years". Setup: pick a sound from the buyer's prime decade.
7. **A/B location, gender, and format of the talking head.** The same copy was tested as a woman in the kitchen (a consistent winner), a park backdrop, a pool, a man, a podcast-mic setup, and different camera depths (chest-up, close-up face). Setup: change only one production variable per test.
8. **Break monotony with edits.** Never run a single-frame talking-head ad: cut between the face, B-roll stock clips, zoom in/out, and horizontal or vertical split-screens; Jairet estimates a static one-frame speaker leaves at least 20% conversion on the table. Setup: keep every on-camera script moving with visual depth changes.
9. **Validate offers against funnel benchmarks before scaling.** New clients must already be validated: VSL retention at 30s / 1 min / 2 min, review-section ("pit") clicks, and order-form conversion all checked against internal benchmarks; an offer far from benchmark is still fine if the ROAS target is met at a high ticket always find the explanation. Setup: benchmark first, cherry-pick the best market idea, test with a few thousand dollars.
10. **Read ad health, then the funnel.** Ad signals like a 5% CTR and low CPC with a page bounce, no play click, or a 3-minute retention cliff point to the funnel, not the ad. Setup: watch bounce-before-play and retention cliffs; fix the funnel, retest with the same ads.
11. **Cut first, rewrite later.** A retention cliff at ~8 minutes was fixed by cutting that segment cleanly (keeping continuity) before any rewriting; A/B it, then iterate. Setup: your VSL player's retention tab is where you find the dead section.
12. **Go mobile-first with the VSL.** If the VSL sits at 16:9, test a 4:5 version with a blurred-fill background behind the video, then a full-bleed 4:5 edit, then a 9:16 full-screen cut; vertical and 4:5 consistently beat 1:1 and 16:9. Setup: re-export the same master in the three aspect ratios.
13. **Strip the sales page to the video only.** Remove the expert bio, bullets, and description under the VSL, keeping only the video plus the compliance footer (disclaimer, terms, privacy); a doctor-expert page cut to image + name went from 15% to 18% conversion, and the headline moved up into the thumbnail. Setup: A/B the page content versus video-only.
14. **Move the headline to the thumbnail and test leads.** With the page stripped, the same headline lives in the VSL thumbnail; 2-3 test leads get written by the client, a freelancer, or TNT when the data shows a lead is the problem. Setup: keep on-page elements down to video + compliance only.

## 4. Tools & Stack
| Tool | Role | Notes |
|---|---|---|
| Trello + ClickUp | Creative production pipeline | Copy due-dates, review stages, hand-offs to editors/actors |
| Google Analytics + native ad analytics | Funnel and ad performance view | Retention, bounce-before-play, CTR/CPC, order-form conversion |
| VTurb / Vurb-style VSL hosting | Video delivery + retention analytics | Retention tabs at 30s / 1min / 2min feed funnel benchmarks |
| A/B testing calculator | Test relevance gating | 95-99% statistical relevance; ~1 week for low volume |
| Google/YouTube + Meta (Facebook/Instagram) ad accounts | Traffic and test platforms | YouTube: 3-5 days per read; Meta: often 1 day, prefer 3 |
| Video editor + stock clip libraries | B-roll, split-screen, zoom edits | Keeps single-frame ads moving; sound effects per test |
| Copy masterminds + swipe/transcription tools | Writing skill + ideation | Copy Accelerator, Nothing Held Back, Fast Forward; transcribe competitor ads |

## 5. Actionable Checklist
- [ ] Build a swipe file of running competitor ads and transcribe what works
- [ ] Write scripts as a two-column doc (copy left, visuals right) for your editor
- [ ] Score every VSL section against its Makepeace-style objective
- [ ] Run 5 new hooks against a winning ad with identical body copy
- [ ] Test 5 sound variants in the first 3 seconds (era-matched noise for older buyers)
- [ ] Put a human face in the first 3 seconds, optionally split-screen with the eye-catching clip
- [ ] Re-edit single-frame talking-head ads with cuts, zooms, and split-screens
- [ ] Prepare 10+ variations of any winner before its fatigue curve starts
- [ ] Benchmark VSL retention and order-form conversion before scaling spend
- [ ] Fix broken funnels with cut-first edits (watch for retention cliffs)
- [ ] Re-export VSL as 4:5 blurred-pillar, full-bleed 4:5, then 9:16 and winner-test
- [ ] Strip the sales page to video + compliance footer and move the headline to the thumbnail

## Post notes
- Source: pt-BR transcript "Segredos da Escala 041" (Jairet Crum, TNT Media). The 30,169-word file contains the episode twice: the full episode runs words 0 to ~15,200 (outro), then replays. Consumed windows 0-6,000 / 6,000-12,000 / 12,000-18,000, plus a tail sample 24,000-30,000 confirming the second copy adds no new content. Rendered to English (host and guest spoke Portuguese).
- Unverified operator claims: TNT "nearly 80 employees"; company "10x since 2019"; the single-ad scale case ($10k $200k/week in 5 weeks; ~$6-7k/day on that ad; ~$800k/month on one ad); "at least +20%" for cutting a single-frame ad; the 15% 18% page-strip case; a 15% "cut-first" VSL gain; ~$2M. none shared as exact client revenue. Contact given in the episode: firstname@tntmedia.com.
- Title number read from the file name ("Segredos da Escala 041" SDE 041); filename carries a ClickBank title, but the episode content is ad/VSL scaling, not ClickBank.