Frameworks: Evolve × Crafter, distilled

Two courses are the reference base for every creative framework in this system:

• Evolve (Spencer / EZ Media, with Shaun & Carlo) — the master reference. Synthesis on the VPS: /srv/mrr/evolve.html. When in doubt about structure, angles, variations or iterations: "base it on Evolve".
• Crafter Club (Nick / Creative Crafters, French course) — the execution and iteration toolkit. Extractions: /root/workspace/adn-crafter/strategy.md, creative.md, execution.md, missing-lessons.md; transcripts /root/workspace/crafter/txt_api/ (4 lessons were badly downloaded and are missing: creative structure, quick wins, TrendTrack research 1 and 2).

Both agree on the core: sell a desire, not a product; niche the message, not the product; answer the market's sophistication; change one thing at a time; micro-details are not tests. Crafter credits Evolve for the sub-desire idea.

---

1. Evolve — the Avatar OS

The chain: PRODUCT → DESIRES → MARKET SOPHISTICATION → CORE AVATAR → SUB-AVATARS → ANGLES → ADS.

1.1 Product (step 0)

• Every product is 2 products: physical (what it is) and functional (what it does). People buy what it does.
• Mechanism = HOW the product delivers the result, not what it does ("supplement that makes you lose weight" → "peptide 47D blocks sugar absorption, so the body burns fat").
• Product Overview Document before any test: name + price · how it works (6th grade) · 5+ features · assumed benefits · assumed desires · assumed new mechanisms (USP) · hidden mechanisms.
• Error #1: targeting desires the product can't honour.

1.2 Desires (Schwartz)

• You can't create desire, only channel it.
• 6 core desires (2026 power ranking): Health/Survival · Status · Sex/Relationships · Comfort · Control · Belonging. Plus mass technological problems (complexity, overwhelm, fragility, maintenance…).
• Surface desires ("I want better sleep") are what avatars are built on; always know the core desire feeding them.
• Evaluate a desire on scope (how many), urgency (now?), staying power (recurring vs one-shot).
• Elevation strategy: move the product up the desire chain to open less saturated markets (stay authentic).
• Desire calendar: Sept structure/fresh start · Oct identity · Nov gratitude + BF · Dec connection · Jan health/control/status (+35–40 % supplement spend) · Feb Valentine's · Mar–Apr renewal · May–Jun achievement · Jul–Aug back-to-school. If performance drops, check where seasonal desires went.

1.3 Market sophistication (5 stages)

99 % of e-com is in stage 3–5. Three responses: new mechanism (most powerful; communicate simply), new information (most accessible: educate on what they don't know → authority → trust; never invent, never fear-monger; "would you run this ad to your mom?"), new identity (hardest; skip for 99 %).

1.4 Avatars

• Goal = RELATABILITY: "That's me" → trust → buy. "Fuck demographics."
• 5 categories, strongest first: desires · experiences · emotions · behaviours/habits · demographics (last).
• Core avatar = one category (one desire). Sub-avatar = core + 1–3 layered categories.
• One desire per avatar (two = soft avatar). The broader the avatar, the better the product/offer/skill must be.
• Revenue stage → number of core avatars: 0–100k/month: 1 (depth in sub-avatars, pick a desire a competitor scales, 2–4 weeks min) · 100–500k: 1, execution · 500k–1M: 2–3 · 1–10M: 3–10 + mechanism
  • info · 10M+: volume monster.

1.5 Desire hunting (research)

Sources in order: Reddit (most honest focus group) · YouTube videos + comments · TikTok comments · Amazon reviews (yours + competitors) · post-purchase survey · AnswerThePublic. Sheet with 3 columns: Desire (I want X) · Customer language (verbatim, never paraphrased) · Ad ideas. "Your research works more on you than you work on it." Never delegate 100 % to AI; always fact-check.

1.6 Creative strategy vocabulary

• Concept (swing): the big idea tested, centred on YOU ("test UGC", "test us-vs-them", "iterate a winner").
• Angle: the reason to buy, centred on the CUSTOMER. "Us-vs-them of WHAT?" — the WHAT is the angle. Every ad needs one; always write it down.
• Variation: a different way to communicate the angle.
• Format: how the ad is delivered. Why an ad flops: no emotional impact · no curiosity · no payoff · too complex (aim grade 5–6) · empty, bossy or repetitive claims · written not spoken · written for you not the avatar · angle too broad · wrong format. The easier the format (image), the better the messaging must be.

1.7 Copy rules

Customer lens only · write like people talk · slippery slide · research = 80 % (2–3 h per hour of writing) · value equation · Kyle's frames (new & only, easy & anybody, safe & predictable, big & fast) · create your own category · show, don't tell · body copy = objection → claim → proof → benefit · awareness levels: never skip steps.

1.8 Testing — the three "guns"

Workflow: Marksman → Sniper → Marksman again at plateau. Volume = manpower: ~1 concept/day, 15–20 ads/week. Feedback loop: identify the sub-avatar/angle/desire behind a winner — never "it's a video so it works".

1.9 Iterating a winner — the 6-axis matrix

1. Length (15/30/45 s) · 2. Gender · 3. Filming location ("where else would people use this?") ·
2. Actor age · 5. Ethnicity · 6. New ways to say the same thing (filmed as real new creatives). Principle: change the execution, keep the angle + winning structure.

1.10 Hooks & scripts

Hook = first 3 s: call out the right audience + promise a benefit + make them want more. Higher in the funnel (unaware) = more scalable. Script = 3 hooks + 3 bridges (~8 s) + 1 universal hold (objection / claim / proof / benefit + CTA), ~43–45 s. Draft → reinforce → refine (read aloud) → visuals → reflect. Always validate claims.

1.11 13 winning static templates (Evolve's Canva set)

Avatar callout · solution exaggeration · stealing credibility · specific pain · accidental ideal outcome · relatable hook · indirect avatar · free/offer · simplified us-vs-them (Nutella) · puberty style · minimal us-vs-them.

1.12 Media buying context (for judging "winners")

One CBO "arena" campaign; if Meta spends on an ad, it's a winner 9 times out of 10; spend IS signal #1. 3-2-2 per ad set (3 creatives, 2 copies, 2 headlines; one concept per ad set). Test 3–7 days, never kill before 72 h. Verdicts: loser · KPI winner · spend winner · breakthrough · super winner. Soft metrics (hook ≥ 25–30 %, hold ≈ 8 %, CTR ≥ 1 %) point to levers; only ROAS/CPA decide. Ad name/batch number on every ad set and ad. Growth Guide: ad type = iteration (mine) / imitation (others') / ideation (my research).

1.13 Terminal errors

Purely demographic avatar · attacking a stage-5 tier-1 market with a mechanism hook · testing one desire for 5 days then pivoting (min 2–4 weeks + 5 sub-avatars) · two desires in one avatar.

---

2. Crafter — the execution toolkit

2.1 Primary desires (13, after Reiss)

Acceptance · Curiosity (pattern breaker: forbid what everyone does) · Family · Idealism (prove it) · Independence · Order · Physical activity (guilt) · Power/money (Nick's favourite) · Romance · Saving (express as an object: "save a Rolex a year") · Social contact · Tranquility · Revenge (a person, or the industry's common enemy). Every creative: 1. name the desire, 2. make it felt (show it, exaggerate), 3. offer a credible mechanism.

2.2 Sub-desire and exclusion

Narrower desire = narrower audience. "Be a big fish in a small pond." Niche creatives, funnel and LP together; the product can stay mass-market. Build MOFU/BOFU creatives on the same niche desire. 4 exclusion levers: who / what / when (trigger) / why (cause + motivation). Niching resets sophistication, raises relevance, lowers the cost of attention. "I've almost never seen a creative too targeted."

2.3 Sophistication ladder (5 rungs) — play N or N+1, never N−1

1 claim · 2 bigger claim · 3 new mechanism · 4 better mechanism · 5 identity/community. Diagnose from competitors' top-10 scaling creatives (simple claims = low; long mechanism VSLs = 3–4; pure branding = 5). Sophistication is per audience; exclusion is the cheapest reset. Never over-sophisticate a virgin market.

2.4 Mechanism = Freshness × Belief; UMP + UMS

Three differentiation cases: format transplant (easiest) · desire transplant · core swap. Inspiration by case: same desire → copy how they market the desire; same mechanism → copy how they explain it; same desire + mechanism, other format → say why yours is better, use their negative reviews. "You don't create belief, you borrow it." Go with existing beliefs, dismantle wrong ones one by one.

2.5 Positioning

The prospect stands in a supermarket aisle full of the same promises. UMP: why the problem happens AND why everything they tried failed, in detail. UMS: "A + B = C". Simplify for a 14-year-old. Map competitors on the same desire (TrendTrack "similar shops"), study their top 4–5 winners, pick the edge (mechanism, authority, narrower sub-avatar). Think whole funnel. "E-commerce is hope."

2.6 Authority (7 figures) — authority validates the MECHANISM, not the product

Country/culture · celebrity · expert · the one who embodies the result · institution · peer · the desired one (the person the buyer wants to please: a pretty woman saying "the best T-shirt for men"; a daughter buying for her dad). Coherence: a wrong authority figure destroys the whole argument. No deepfakes, no invented studies.

2.7 Data: internal vs external; golden nuggets

Internal (ad comments, BM stats, support tickets, objections, reviews, post-purchase surveys) = higher hit rate. External (competitors, trends, persona conversations, out-of-niche brands) = big swings. Golden nugget = a review/comment that nails you to the wall: validated by many, the target recognises itself totally, an instant click. Wording sits under the angle (angle "grow your hair"; wording "receding temples"). When a new wording wins, re-run all past winners with it ("cascade"). "Your customer is your best copywriter." "The customer must feel we stole their diary."

2.8 The creative equation

Creative = Format + Angle (the outcome, not the problem) + Persona + Product + Hook (+ funnel stage), written into the ad name. Sort BM by spend, decompose the top creatives, change one variable at a time, keep a table. Never iterate micro-details. A headline change that keeps the angle is a valid iteration.

2.9 Value equation (Hormozi applied)

Value = (Dream outcome × Perceived likelihood) / (Time delay × Effort). Show every component on screen. Don't cut delay/effort so far that likelihood collapses. Likelihood raisers: authority, debunking what fails, staged timelines, guarantee, native well-acted execution, visible mechanism, hyper-specific persona, naming the medication they already take.

2.10 Script blocks, hooks, CTA

Base: Hook → Problem → False solution → Product intro → Features → CTA. Classic brief: Hook → restate problem → solution → product in use → social proof (always, many faces) → angle-specific objections → CTA. Structure depends on awareness. One script = one angle. More promises = less credibility. Zero repetition. Hook = targeting; don't optimise hook rate. 19 hook types (see copy-analysis.md). One CTA + one booster. 4 B-rolls always requested.

2.11 Formats and statics

19+ formats (IG-story testimonial, UGC, podcast, news set, comparison, mash-up, unboxing, comment reply, green screen, street interview, founder, tutorial, "3 reasons", POV, interview, 2–3 person discussion, ASMR, silent aesthetic, storytelling). Every video format can become a static and vice versa. "If it doesn't work, it's your angle and your message, not your format." Statics: 80 % of the result is wording; 500 ms test; simplicity scales; never soulless corporate lines; explicit direct benefits; congruence; visual hierarchy ("close your eyes, open them"); a static has a hook. 10 static templates + 11 winning statics decomposed (see copy-analysis.md and template-reading.md).

2.12 Iteration framework

Winner = ROAS > break-even; beginner ≥ €500 spend profitable; agency ≥ 10–15 % of monthly spend (20–25 % = very big). Process: find winners (sort by spend) → decompose (equation) → cross with other proven angles/personas/formats (find the gaps) → produce fast. Variables: angles · formats · actor age · ethnicity · gender (education for women often works better with a male presenter) · location · edit · length (shorter = lower funnel) · other product · hook/headline · persona · funnel stage · language/market · static ↔ video · answer a top comment · isolate a sub-angle. Wording rule: when transposing a winner, keep the same hook/headline ~80 % of the time.

2.13 Scale ("we don't scale budgets, we scale creatives")

6 levers + 1 multiplier: angles · personas/sub-avatars · diversity · iterations · full funnel + congruence (one LP per avatar/angle at scale) · volume; multiplier = a tight data feedback loop. Funnel spend base 60 % ToF/MoF, 25 % awareness 3–4, 15 % most aware; never starve the top. 70–80 % of tests fail. 20–30 creatives before killing a hypothesis. Unaware winners have ~20× the impact.

---

3. Where they disagree — and the house rule

---

4. How they combine — the merged AI-assisted pipeline

TL;DR (Evolve): Product (mechanism) + Desire (surface) + Sub-avatar (1–3 layered categories) + Angle (hook entry point) + Sophistication response = winning ad. The only output that counts: "That's me" → trust → buy.

---

5. Using the frameworks in ANALYSIS (not production)

When decoding a competitor ad, run these lenses in order and write one line each:

1. Desire named / shown / mechanism offered? (Crafter's 3 steps) Which of the 13?
2. Sophistication response used (claim / bigger claim / new mechanism / better mechanism / identity / new information / exclusion)?
3. Awareness where the ad starts and when the product appears (unaware reveals late ~12 s; problem-aware 8–10 s; product-aware opens on the product).
4. Sub-avatar by exclusion (who/what/when/why/tried).
5. Angle as outcome (one sentence).
6. Mechanism UMP/UMS + authority figure type.
7. Value equation: which of the 4 components is on screen; which is missing.
8. Creative equation: format + angle + persona + product + hook + funnel stage → the code you'd name it with.
9. Diagnosis: what lever they are iterating across their variants (that's what they believe in).
10. Gap: the combination (winning format × winning angle) they have NOT made — our first test.

Note on ratios: older notes recommend a fixed 80/20 (iterations/new concepts) or 60/40. The current house method is to clone proven winners and not impose a fixed ratio; use Crafter's flex (hot period → more iterations; nothing working → more net-new) only as a sanity check.