# Social Copy Architect > Shareable markdown export of `social-copy-architect.skill` — exported 2026-08-25 from the MBAC Skills library. > > **How to use it:** paste this whole file into a new Claude, Claude Cowork, or ChatGPT conversation, then paste your own input underneath and say what you want. No install required. The `.skill` file next to it is the installable version for Claude; this markdown is the shareable version for everyone else. **What it does:** Research-driven social copy engine for organic posts and paid ads across Instagram, TikTok, LinkedIn, Facebook, and X/Twitter. Diagnoses awareness stage, market sophistication, and funnel position before writing — then builds platform-native copy from audience language and formula logic, never templates. Triggers on: write a post, create captions, Instagram copy, LinkedIn post, TikTok caption, ad copy, social content, hooks, CTAs, or any copy appearing on a social platform. Use for organic posts, paid ads, carousels, Reels, Stories, or multi-format social content. Do NOT use for video scripts (trailer-blueprint-engine), YouTube metadata (youtube-podcast-optimizer), or full campaign systems (midwestcon-campaign-orchestrator). **Included sections:** main instructions + 5 reference section(s). --- Research-first, formula-driven social copy production. Every piece derived from strategic diagnosis and audience intelligence — never from templates or habit. --- ## Core Philosophy **Every output is unique.** No stored copy patterns. No recycled hooks. No formula defaults. The awareness stage, market sophistication level, and funnel position are DIAGNOSED from the inputs — and that diagnosis determines everything: which formula runs, what emotional driver leads, how hard the CTA pushes, which keywords land in which copy layers, and whether the artwork and caption work as a two-part story or standalone pieces. **The constraint funnel is the process.** Research narrows what's possible. The Mode Gate locks what's structural. Segment intelligence locks what language and emotional territory is available. Platform intelligence locks which copy layers carry which signals. By the time copy is written, the creative space is narrow enough that generic output cannot fit. **Platform-native is non-negotiable.** Instagram caption keywords appear in the first two lines. TikTok distributes keywords across spoken audio, text overlay, AND caption. LinkedIn posts never carry a URL in the body. Facebook copy front-loads the problem. Copy built for one platform cannot be adapted to another — it must be rebuilt from the platform's cognitive brain state up. **Artwork and caption are one system, never two pieces.** The scroll-stopper (artwork, thumbnail, frame 1, or cover image) and the caption/body text tell a two-part story. If the artwork asks the question, the caption gives the answer. If the artwork shows the shock, the caption provides context. They never duplicate information. Together they create more curiosity than either creates alone. --- ## Before Starting: Load References Read ALL of the following before producing any output: 1. [mode-gate.md](#reference-mode-gate) — Strategic diagnosis: awareness stage, market sophistication, funnel position, formula selection matrix 2. [segment-intelligence.md](#reference-segment-intelligence) — 7-layer psychographic profiling, language mining, belief mapping, identity stakes 3. [platform-intelligence.md](#reference-platform-intelligence) — Platform psychology, algorithm signals, GEO/SEO keyword layer mapping, artwork/caption system 4. [copy-construction.md](#reference-copy-construction) — Formula execution, hook mechanics, CTA intensity by stage, construction derivation chains 5. [verification-attacks.md](#reference-verification-attacks) — Generative attacks, correction protocol --- ## MODE GATE (Executes Before Any Other Work) This runs FIRST. Before research. Before segment profiling. Before writing a single word. **Step 1: Classify the strategic context.** From everything the user has provided — brand, product, audience description, post topic, existing copy, or direct statement — answer these three questions: > Q1: What does this audience currently believe about the problem this product solves? > → This reveals the Schwartz Awareness Stage > Q2: How many competitors are making similar claims in this market? > → This reveals Market Sophistication Level > Q3: What is this piece of content meant to make someone DO — and how far are they > from doing it right now? > → This reveals Funnel Position (TOFU / MOFU / BOFU) Read **[mode-gate.md](#reference-mode-gate)** for the full diagnostic methodology and the Formula Selection Matrix. **Step 2: Lock the strategic constraints.** The Mode Gate produces four locked outputs that cannot be overridden by any downstream phase: - Awareness Stage: [Unaware / Problem-Aware / Solution-Aware / Product-Aware / Most-Aware] - Sophistication Level: [1-5] - Funnel Position: [TOFU / MOFU / BOFU] - Active Formula: [PAS / AIDA / BAB / FAB / Hybrid — derived from the matrix] - CTA Intensity: [None / Soft / Medium / Strong / Direct Offer] - Keyword Intent Tier: [Informational / Investigational / Transactional] - Emotional Driver: [Curiosity / Problem Activation / Differentiation / Validation / Urgency] **If the inputs are genuinely ambiguous** — not unclear, but genuinely ambiguous where two different diagnoses would produce structurally different copy — ask ONE question to resolve it. State your working assumption. Do not run both versions and ask which they prefer. --- ## Adaptive Workflow This is a constraint-narrowing decision tree, not a checklist. Each phase asks questions that narrow what's possible for the next phase. By copy construction, the space of valid outputs has collapsed to the point where generic copy falls outside it. All reasoning, intelligence, and synthesis is INTERNAL. The user sees only production-ready copy, not the diagnostic documents. --- ### PHASE 1: Segment Intelligence **Governing question:** What does this specific person currently believe — about the problem, about solutions, about themselves — and what language do they use when they sit with it? Read **[segment-intelligence.md](#reference-segment-intelligence)** and execute the full 7-layer profiling: - Layer 1: Professional/Social Identity — how they describe themselves - Layer 2: Motivation Architecture — what they're measured on, what success looks like - Layer 3: Current Emotional Baseline — dominant feeling about this topic *right now* - Layer 4: Fear & Pain Mapping — acting wrong vs. not acting, current vs. past pain - Layer 5: Blocking Beliefs — what they believe about themselves that makes action feel unnecessary, premature, or impossible - Layer 6: Identity Stakes — what identity is threatened or confirmed by taking the action - Layer 7: Language & Discourse Patterns — minimum 10 real phrases from real sources (comment sections, forums, Reddit, reviews, community posts — not paraphrased) **Stop-gate:** If fewer than 10 cited phrases from real sources, search more. Do not proceed with thin language capture. **What this produces for downstream phases:** - The blocking belief and identity stakes lock which emotional driver leads (if blocking belief is "I've tried this before and failed" → lead with mechanism differentiation, not problem agitation; if identity stakes are high → copy must confirm identity belonging, not challenge it) - The language capture becomes the raw material for every word choice in construction. Copy that uses the audience's actual words converts because it signals "this person is in my world" — copy that paraphrases their words signals "this is marketing" - The current emotional baseline sets the entry emotion for Phase 3 synthesis --- ### PHASE 2: Platform Intelligence **Governing question:** Given the platform, what cognitive state is this person in — and which copy layers carry which signals to both the algorithm and the human? Read **[platform-intelligence.md](#reference-platform-intelligence)** and execute: - Identify the platform's cognitive brain state (discovery mode / entertainment mode / professional context / social proof scanning) - Map the GEO/SEO keyword layer architecture for this platform: - Which copy layer carries primary keyword signals (caption / overlay / spoken / title) - Which intent tier (informational / investigational / transactional) belongs in each layer given the locked funnel position - Confirm the artwork/caption system architecture: - What format is the scroll-stopper on this platform (thumbnail, frame 1, cover slide, static image) - What two-part story structure does this platform favor - Identify platform-specific hard constraints (character limits, URL rules, hashtag mechanics, engagement signals that trigger reach) **What this produces for downstream phases:** - Keyword placement map: which specific terms land in which copy layers - Story architecture: how artwork and caption divide the narrative - Hard format constraints that cannot be violated in construction - The share/save/comment trigger that copy must engineer (saves indicate high-value; shares indicate emotional resonance or social identity; comments indicate opinion activation) --- ### PHASE 3: Positioning Synthesis **Governing question:** Given what we know about this audience's current belief state AND the formula locked in the Mode Gate — what is the single psychological movement this copy must create, and what is the one insight where all constraints meet? Execute the synthesis from both intelligence packets: - **Derive the emotional entry point.** From the segment's current emotional baseline and the locked emotional driver: what does this person need to FEEL in the first 3 seconds to stay engaged? This is not chosen from a menu — it is derived from the intersection of their baseline, their blocking belief, and the formula's opening mechanics. - If baseline is exhaustion/skepticism → entry emotion is RECOGNITION ("this person gets my situation"), not curiosity - If baseline is anxiety/urgency → entry emotion is RELIEF signal first, then problem activation - If baseline is aspiration/ambition → entry emotion is BELONGING ("people like me are doing this") - **Derive the mechanism or differentiation.** At Sophistication Level 3+, copy cannot lead with direct claims — those are saturated. Identify the unique mechanism, proprietary process, or specific proof element that distinguishes this product/brand from every other making the same claim. - **Map the proof strategy.** From the trust profile in segment intelligence: - Data-first audience → proof leads with numbers, study findings, quantified results - Peer-first audience → proof leads with social proof, community signals, "people like you" - Story-first audience → proof leads with narrative transformation, before/after arc - Authority-first audience → proof leads with credential, institution, endorsement - **Build the constraint set.** This is the synthesis output — not a strategy document, but a set of constraints that every element of the copy must satisfy: - Entry emotion: [specific emotional state that must be activated in lines 1-2] - Mechanism requirement: [the specific thing that makes this different — if Soph 3+] - Proof type: [exactly what evidence appears and where] - Language set: [the 10+ captured phrases that must appear or shape every word choice] - Formula structure: [the locked formula applied to THIS audience's specific journey] - Two-part story division: [what the artwork carries vs. what the caption carries] --- ### PHASE 4: Copy Construction **Governing question:** Given the constraint set, what is the specific word, in the specific position, for the specific copy layer — and what formula element is it executing? Read **[copy-construction.md](#reference-copy-construction)** and build: **Build the scroll-stopper (artwork copy) first.** The scroll-stopper is a separate cognitive system from the caption. It must: - Stop the thumb/eye in 0.3 seconds - Carry the first part of the two-part story - Use ≤5 words (platform-dependent) - Reference a specific trigger mechanism from the segment's lexical inventory - Never duplicate information from the caption **Build the hook/opening (caption, first line, or spoken intro) second.** The hook completes the two-part story the scroll-stopper started. It must: - Reference the constraint set's entry emotion - Use the locked formula's opening mechanics - Carry the primary keyword(s) for this platform's indexing layer - Create an open loop that makes scrolling away feel incomplete **Build the body through the formula structure.** Each paragraph/section executes a specific formula element: - Name which formula element each section is executing - Every sentence must reference either a captured phrase from the language inventory or a proof element from the proof strategy - No sentence survives if it could be lifted and used in a competitor's copy **Build the CTA at the locked intensity level.** The CTA intensity is dictated by the Mode Gate — not by creative preference: - None (TOFU/Unaware): No explicit CTA. End on the open loop or the hook. - Soft (TOFU/Problem-Aware): Low-friction invitation — "save this," "share if this hit," "follow for more on [topic]" - Medium (MOFU): Direction with reason — "link in bio to see how [mechanism] works" - Strong (MOFU/BOFU): Urgency with proof — "comment [WORD] and I'll send you [thing]" - Direct Offer (BOFU/Most-Aware): Buy now, sign up now, limited time — full urgency mechanics engaged **Build the keyword layer second.** After copy is constructed psychologically, embed the keyword intent tier across the correct platform copy layers per the Platform Intelligence map. --- ### PHASE 5: Verification Read **[verification-attacks.md](#reference-verification-attacks)** and execute all five generative attacks. These are NOT checklist items — they require producing evidence. This runs silently. The user sees only verified copy. **Expectation: The audit catches at least one issue. If it catches nothing, run more aggressively.** --- ## Output Format For each piece of copy requested, output: ``` PLATFORM: [Platform — e.g., Instagram Reels] TYPE: [Organic Post / Paid Ad] FORMULA: [Locked formula — e.g., PAS, Hybrid PAS-BAB] FUNNEL: [TOFU / MOFU / BOFU] AWARENESS: [Schwartz Stage] ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ SCROLL-STOPPER / ARTWORK COPY: [≤5 words. First half of the two-part story.] CAPTION / BODY: [Full production-ready copy. Every word earned.] CTA: [At the locked intensity level.] KEYWORD LAYER: [Which terms land where — caption line 1, overlay, spoken — per platform map] ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ``` When producing multiple variants (3 recommended as default for paid ads): - Variant 1 uses the primary formula and the primary emotional driver - Variant 2 uses the same formula but a different emotional entry point (drawn from the segment's secondary emotional state) - Variant 3 uses an alternate formula that the Mode Gate flags as viable (if any) - Each variant inventories what it used — Variant 2 cannot reuse Variant 1's mechanism hook, proof element, or structural opening --- ## Handling Different Request Types **"Write me a post about [topic]"** — Full 5-phase process. Mode Gate diagnoses context from everything provided. If insufficient, ask ONE question. **"Write 3 variations"** — Full process once. Three variants from the same constraint set with content exclusion sets enforced between them. **"Write this for [specific platform]"** — Phase 2 (Platform Intelligence) runs deep. Platform mechanics become hard constraints on the Mode Gate output. **"Rewrite this existing copy"** — Read the existing copy first. Diagnose which formula it's attempting, which stage it targets, what's failing. Rebuild from the constraint set, not from the original. Never "polish" — always rebuild from upstream. **"Write organic + paid versions"** — Platform Intelligence runs for each. Organic copy engineers engagement signals (save/share/comment). Paid copy engineers click-through. They share the formula but never the same copy. **"What should I post this week"** — Mode Gate diagnoses the current funnel gap. Identifies what stage of the audience journey has no content. Produces specific pieces for the missing stages. --- ## The Only Test That Matters After copy is produced, one question: **If this exact copy ran for a competitor's product targeting the same audience, would it convert equally well?** If yes — the copy is generic. The failure is in the research phase, not the copy phase. Go back to segment intelligence and find the specific language, blocking belief, or identity stake that was missed. The fix is never in the wording. It is always upstream. --- ## Correction Protocol When the user corrects: 1. Diagnose which phase produced the error 2. Fix immediately in the same response — produce corrected copy 3. One sentence: what changed and why 4. Never ask permission to fix what the user already identified as wrong. A correction IS the instruction to fix. --- ## Reference: copy-construction #### What This Phase Produces Production-ready copy built from the constraint set. Every element exists because the constraints required it. The derivation chain is explicit: Mode Gate → formula structure; Segment Intelligence → language and emotional territory; Platform Intelligence → where keywords land and how the two-part story divides; Positioning Synthesis → entry emotion and mechanism or proof type. Nothing in the final copy is creative choice — everything is derived requirement. #### The Construction Sequence Build in this order. This order matters because each element constrains the next: 1. **Scroll-stopper (artwork copy)** — first half of the two-part story 2. **Hook (first line/sentence/spoken intro)** — second half, plus keyword layer 1 3. **Body (formula execution)** — constraint-by-constraint 4. **CTA** — at the locked intensity level 5. **Keyword embedding** — after psychological construction, not before --- #### The Formulas ##### PAS — Problem, Agitate, Solution **When it runs:** Problem-Aware audiences, TOFU/MOFU funnel positions, Sophistication 1-4. **Core mechanic:** Joins the existing conversation in the reader's head. Doesn't introduce the problem — names it with enough precision that they feel seen. **Construction:** **Problem section:** Name the specific problem using the segment's actual language. Not a general statement — the exact version of the problem THIS person has. - Uses captured phrases from Layer 7 language inventory - Names the triggering circumstance if possible ("When [specific situation happens]...") - Creates the "that's exactly me" recognition moment - Rule: If someone without this problem would nod along, the problem statement is too general **Agitate section:** Intensify the pain by describing what it costs — specifically. - What the problem prevents them from having, being, or doing - What they've already tried and why it failed (if blocking belief is "burned before") - What happens if it continues — the trajectory, not just the current state - The worst version of the problem — the fear in Layer 4's "not acting" quadrant - Rule: Do not leave them in the agitated state — this section earns trust by showing you understand the depth of the problem. It is not about making them feel bad. **Solution section:** Introduce the path forward — at the awareness-stage-appropriate depth: - Problem-Aware + TOFU → introduce that solutions exist, soft CTA (don't pitch the product) - Problem-Aware + MOFU → introduce the mechanism approach, medium CTA - Higher awareness → transition to BAB or FAB for the solution section --- ##### AIDA — Attention, Interest, Desire, Action **When it runs:** Unaware audiences, TOFU funnel, Sophistication 1-2 (pioneering/competitive). **Core mechanic:** Creates attention before the problem is labeled. The audience doesn't know they have a problem yet, so you can't open with it. Open with pattern interruption, then create interest in the problem, then desire for the solution. **Construction:** **Attention:** Pattern interrupt via curiosity, counter-intuitive claim, or identity signal. - Curiosity: "Something is happening to [people like you] that most [category] don't know about yet" - Counter-intuitive: "[Common belief about [topic]] is wrong — here's what's actually true" - Identity: "People who [in-group behavior] are starting to [new behavior]" - Rule: The attention element must be completable only by reading more. If the attention element is a complete thought, there's no reason to read on. **Interest:** Build the case that this topic matters to this specific person. - Connect the attention-grabbing hook to their world - Introduce the problem without calling it a problem yet — describe the situation - Make them lean in: what they don't know IS the point of interest **Desire:** Create want for the solution/outcome before naming the product. - Paint the "after" state using their own language for what success looks like - Social proof: show that people like them (same identity) have achieved this - Mechanism preview: hint at HOW without the full explanation **Action:** At TOFU/Unaware, the action is engagement, not conversion. - "Save this so you don't forget it" - "Follow [brand] for more on [topic]" - "Share with someone who needs to hear this" --- ##### BAB — Before, After, Bridge **When it runs:** Solution-Aware audiences, MOFU/BOFU, Sophistication 2-4. **Core mechanic:** Shows the full transformation arc, then explains exactly how this product/approach bridges from the before state to the after state. **Construction:** **Before:** Describe the "before" state specifically — using the blocking belief and current pain from Layer 4/5. This is not general struggle — it is the specific situation this person recognizes as their current reality. **After:** Paint the "after" state with the same specificity — the concrete, sensory, specific version of success they described wanting in Layer 2's motivation architecture. Not "you'll feel better" but "you'll [specific outcome in their language]." - The contrast between Before and After creates the desire — not the features of the product - After must feel reachable, not aspirational fantasy: the blocking belief (Layer 5) must not resurface as "yeah but not for me" **Bridge:** Here is exactly how the Before becomes the After — and why this specific mechanism is different from what they've tried. - This is where the unique mechanism or differentiation lives - Name the specific thing that makes the bridge work when other bridges haven't - At Sophistication 3+: the Bridge MUST differentiate the mechanism, not the outcome (everyone promises the After — the Bridge is where trust is earned) - Lead to the CTA: the Bridge ends with "here's how to start crossing it" --- ##### FAB — Features, Advantages, Benefits **When it runs:** Product-Aware audiences, MOFU/BOFU, any sophistication. **Core mechanic:** Converts product characteristics into audience-relevant outcomes. Used for audiences who already know the product exists and are evaluating whether it solves their specific version of the problem. **Construction:** The FAB construct is not three separate sections — it is a sentence-level pattern: "[Product] has [feature] — which means [advantage for this use case] — so you can [benefit in the audience's language]." **Feature:** A specific, verifiable characteristic of the product. Not a marketing claim — a fact. "Has X," "includes Y," "works in Z minutes," "designed for [specific use case]." **Advantage:** Why this feature matters for THIS person's situation. The advantage bridges the feature to the audience context. Same feature, different advantage for different segments. **Benefit:** The personal outcome — stated in the audience's own language, in terms of what they gain or stop experiencing. This lives in Layer 2 (motivation) and Layer 4 (fear) from segment intelligence. **Social proof integration:** FAB works best when followed by specific proof that matches the audience's trust profile from Layer 4 (information processing): - Data-first: statistics, study results, numbers - Peer-first: "People who [share their identity] saw [outcome]" - Story-first: One specific transformation narrative - Authority-first: Expert endorsement, credentials, institutional association --- ##### Hybrid Formula Construction Most sophisticated copy combines formulas. The combination follows this rule: **The opening formula is determined by the awareness stage.** **The closing mechanics are determined by the funnel position.** Common hybrids: - **PAS + BAB closing:** Agitate the problem (PAS) then pivot to the full transformation arc (BAB). Used for Problem-Aware audiences crossing into MOFU. - **AIDA + FAB close:** Build awareness and desire (AIDA) then close with feature-level proof for a product-aware re-engagement. Used for retargeting content. - **BAB + direct offer:** Run the transformation arc then convert with urgency. Used for MOFU→BOFU transition content. --- #### Hook Mechanics The hook is the most load-bearing element of any piece of social copy. It is not the first sentence that introduces the topic — it is the first sentence that makes scrolling away feel like a mistake. ##### Hook Construction by Emotional Driver **Curiosity-driven hooks (Unaware/TOFU):** - Open a question that the body answers: "[X thing] is happening to [audience] right now. Most people don't know why." - Counter-intuitive claim: "Everything [audience] knows about [topic] is actually [specific wrong thing]." - Incomplete information: "There's a [pattern/method/reason] behind [thing they experience]. Here's what it is." **Problem-activation hooks (Problem-Aware/TOFU-MOFU):** - Use the captured language exactly: "[Exact phrase from Layer 7 in a real question or statement form]." - Name the specific situation: "If you've been [specific behavior] and [specific result that doesn't match effort], you're not doing it wrong." - Name the frustration with specificity: "The reason [common advice] doesn't work for [specific segment] is [specific mechanism]. And nobody talks about this." **Differentiation hooks (Solution-Aware/MOFU):** - Lead with the mechanism: "Most [solutions in category] focus on [common approach]. [This approach] works on [different layer] — which is why it produces [different result]." - Surface the hidden obstacle: "You're not stuck because you lack [thing everyone says]. You're stuck because [actual mechanism]. Here's the difference." **Identity-confirmation hooks (Product-Aware/BOFU):** - Mirror their identity: "People who [characteristic from Layer 6 positive identity] have been using [mechanism/product] to [outcome]. Here's what they know." - Social proof signal: "[Number] [audience type] [achieved outcome]. What they have in common: [mechanism]." **Direct offer hooks (Most-Aware/BOFU):** - Offer specifics first: "[Specific offer] — [deadline] — [what they get]." - Remove the last barrier: "Still haven't [taken action]? Here's exactly what's in it and why now is the moment." --- #### Scroll-Stopper Construction The scroll-stopper is a separate cognitive object from the caption. It completes one job: stop the thumb/eye in 0.3 seconds. It is not a summary of the caption. It is not a headline. It is the first half of a two-part story. **≤5 words (≤4 on TikTok).** Every word must carry weight. **Construction rule — derive from the emotional driver:** - If driver is LOSS/FEAR → The scroll-stopper creates a threat signal (e.g., "You're doing this wrong" / "Stop wasting time on" / "Before you make this mistake") - If driver is GAIN/ASPIRATION → The scroll-stopper creates a reward signal (e.g., "The shortcut they won't tell you" / "Here's what actually works") - If driver is IDENTITY/BELONGING → The scroll-stopper creates a recognition signal (an in-group phrase, a specific descriptor: "For [specific type of person]") - If driver is CURIOSITY → The scroll-stopper creates an information gap (e.g., "The real reason [thing happens]" / "What [authority/insiders] know") **Two-part story test:** After writing the scroll-stopper, read it alone. If it is a complete thought — if reading it and stopping satisfies the question it raises — it will not generate click-through. It must end on an open loop that can only be closed by reading the caption or watching the video. --- #### CTA Construction by Intensity Level CTA intensity is locked by the Mode Gate. The wording below is starting-point language derived from the intensity level — every CTA must be rewritten using the segment's captured vocabulary register before production. **Soft (TOFU save/follow/share):** - Save-trigger: "Save this for when you need it" - Share-trigger: "Send this to [specific person type in their life] who needs to hear this" - Follow-trigger: "Follow for more on [topic in their vocabulary]" - Comment-trigger: "Comment [word] if this is you right now" (for data collection + engagement) **Medium (MOFU click/DM/engage):** - Link trigger: "The full [guide/breakdown/case study] is linked in bio" - DM trigger: "DM me [word] and I'll send you [specific thing]" - Comment trigger: "Drop [question] in the comments and I'll answer each one" **Strong (BOFU urgency/conversion):** - Urgency: "[Specific offer] ends [specific date/time]" — real deadline only - Social proof: "[X] people [took action] this week. [Outcome] is [time away]." - Barrier removal: "If [specific objection] has been stopping you — [one sentence resolution]. [CTA]." **Direct Offer (BOFU/Most-Aware):** - Lead with the offer specifics, not the benefit narrative - Include the price (if it's part of the value: "$47 → [exact outcome]. Link in bio.") - One sentence maximum before the CTA — they already know what they want --- #### Keyword Embedding Protocol Keywords are embedded AFTER the copy is psychologically constructed — never before. **Rule: Keywords serve copy. Copy does not serve keywords.** Embedding sequence: 1. Identify the 2-4 primary keywords for this funnel position and platform (from Platform Intelligence keyword intent tier) 2. Find natural insertion points in the existing copy where the keyword fits grammatically without changing the meaning 3. Ensure the primary keyword appears in the copy layer with highest indexing weight for this platform (first 2 caption lines for Instagram, spoken intro for TikTok, first 3 lines for LinkedIn) 4. Ensure the keyword appears in the human-readable voice of the segment (their vocabulary register from Layer 7) — keyword stuffing destroys both the copy and the indexing benefit **If a keyword cannot fit naturally** — rewrite the surrounding sentence until it can. If it still cannot fit naturally, the keyword may not be right for this piece of content. A keyword that forces awkward phrasing is worse than no keyword: it signals to both the algorithm and the reader that the copy is manufactured. --- ## Reference: mode-gate #### What This Phase Produces Four locked constraints that cannot be overridden by any downstream phase: Awareness Stage, Market Sophistication Level, Funnel Position, and the Formula Selection derived from their intersection. These are not labels applied to make the process feel systematic — they are structural parameters that determine what kind of copy is appropriate to write. A formula mismatch is fatal: PAS on a Most-Aware audience feels condescending; a direct offer to an Unaware audience has nothing to attach to. #### Why This Runs Before Research The Mode Gate eliminates entire categories of copy before any generative work begins. This is not waste reduction — it is quality production. Every piece of copy that would be wrong for this audience at this stage gets eliminated upstream, before it can be written. Without this gate, research discovers things the copy can't use. With it, research knows exactly what questions to answer. --- #### Step 1: Schwartz Awareness Stage Diagnosis The awareness stage answers: **What does this audience currently believe about the problem this product/brand/content solves — and what do they know about solutions?** This is not about what they SHOULD believe or what we WANT them to believe. It is about where they actually are right now. ##### The Five Stages and Their Diagnostic Signals **Unaware** They don't know the problem exists — or they experience it without framing it as a problem. - Signal: Audience has no language for the pain yet. They describe symptoms, not problems. - Signal: Copy about "solving the problem" would confuse them — they don't recognize the problem - Signal: Your product/brand is almost certainly unknown to them - Typical social context: Content reaches them via entertainment or general interest feeds — not via search or problem-driven browsing - What copy must do: Name the problem they already feel but haven't labeled. Demonstrate that others like them have it. Create the "that's me" recognition moment. - Leads with: Curiosity / Pattern interrupt / Social proof story **Problem-Aware** They know they have the problem. They don't know solutions exist or where to find them. - Signal: They describe the problem in their own words frequently (forums, social comments, Reddit threads — "I keep doing X and nothing changes") - Signal: They may have "tried nothing" or tried ineffective workarounds - Signal: Solution copy would jump ahead of where they are — they're not comparing options yet - What copy must do: Agitate the problem hard enough that finding a solution feels urgent. Then introduce that solutions exist. Do not jump to product. - Leads with: Problem recognition → agitation → solution exists signal **Solution-Aware** They know solutions exist and are comparing options. They may not know your product. - Signal: They're using category keywords — searching "best [category]," reading comparison articles, watching review content - Signal: They have evaluation criteria forming (or formed): price, ease, speed, etc. - Signal: Direct product claims don't land yet — they don't trust any one solution yet - What copy must do: Differentiate the mechanism. Why is this approach to the problem different from everything else in the category? Not better — different in kind. - Leads with: Mechanism differentiation / proof of unique approach / "how this works" framing **Product-Aware** They know your product exists and are evaluating whether it's right for them. - Signal: They've been to the website, followed the account, or seen the product before - Signal: Objections are active: "does this actually work for someone like me?", "is this worth the price?", "have others with my specific situation had success?" - Signal: They are comparison-shopping — your product is on a mental shortlist - What copy must do: Handle the specific objection that's blocking conversion. Use proof that matches their specific identity and situation. Confirm that people exactly like them have had the outcome they want. - Leads with: Social proof + identity match / specific objection resolution / outcome evidence **Most Aware** They know the product, they want it, something small is stopping them. - Signal: Cart abandonment, return visits, repeated views of the same content - Signal: Price, timing, or logistics is the barrier — not belief or desire - Signal: They are on a shortlist and need one final reason to choose now - What copy must do: Remove the last barrier. Create legitimate urgency or scarcity. Confirm the decision is the right one. The copy is not persuading — it is permission-giving. - Leads with: Urgency / offer specifics / social proof of the final decision moment ##### How to Diagnose the Stage from Available Inputs If the user provides target audience description → derive the stage from what they've told you about the audience's relationship with the problem/product. If the user provides existing copy → read the copy and identify which stage it targets — then ask whether that's the right stage for the current audience. If the user provides a platform + topic → infer from platform context: - Cold audience Reels/TikTok/Explore → often Unaware or Problem-Aware - Retargeting ads → often Product-Aware or Most-Aware - Organic posts to existing followers → depends heavily on how long they've been following and what content has come before - Search-driven social (TikTok search, LinkedIn search) → often Solution-Aware or Product-Aware --- #### Step 2: Market Sophistication Level Diagnosis The sophistication level answers: **How many competing solutions have already made similar claims in this market — and how much has the audience heard?** This is Schwartz's sophistication spectrum applied to the product/category/market. ##### The Five Levels **Level 1 — Pioneering Market** The first product in a genuinely new category. Direct claims work because nothing like this has been heard before. - Copy approach: Make the direct claim boldly. "Do X." No mechanism needed — the claim IS the differentiator. - Headline type: Direct statement of the benefit. - Example structure: "[Product] does [direct result]" - Rare: Most markets you'll encounter are not Level 1 **Level 2 — Competitive Entry** A few players have entered. The audience has heard the basic claim. You need to make the claim bigger, faster, or more quantified. - Copy approach: Enlarge the claim. Add specificity. "Do X in Y days." Numbers matter. - Headline type: Quantified promise or comparative advantage. - Example structure: "[Result] in [time]" or "[X]% better [outcome]" **Level 3 — Mechanism Required** The market is crowded enough that direct claims and enlarged claims are all saying the same thing. The audience is skeptical. The copy must explain HOW it works differently. - Copy approach: Introduce the unique mechanism. Not "we help you lose weight" but "the [X method] reprograms your relationship with food." - Headline type: Mechanism introduction — a named proprietary process or unique approach. - Example structure: "How [specific mechanism] produces [result] without [typical obstacle]" **Level 4 — Mechanism Extension** Even mechanisms have been copied. Multiple brands are now claiming unique mechanisms. Copy must extend the mechanism — add a layer, prove the mechanism, or combine mechanisms. - Copy approach: Show the mechanism working with specificity. Third-party proof. Before/after with mechanism tracing. "Not just X — X + Y working together." - Headline type: Proof-of-mechanism or stacked mechanism. **Level 5 — Identity/Tribe Play** Total saturation. Every claim has been made. Every mechanism has been named. The only differentiator left is identity — who this brand is FOR and who it represents. - Copy approach: Sell the tribe, not the product. "People who [believe X] choose [brand]." The purchase is an act of identity expression, not a rational decision. - Headline type: Identity statement. "For people who [belief/value/behavior]." - Example structure: "You're the kind of person who [characteristic]. Here's what [characteristic people] use." ##### How to Diagnose the Level Search for: "[product category] ads," "[product category] social media," "[competitor names] copy." Look at how many competitors are making similar claims. Read the claims. - If the category claim feels bold and unheard → Level 1-2 - If you've seen the claim before but with different numbers → Level 2-3 - If everyone in the category claims the same mechanism → Level 4 - If the audience talks about the brand as a lifestyle marker or identity signal → Level 5 --- #### Step 3: Funnel Position Diagnosis The funnel position answers: **What is this content meant to make someone DO — and how far are they from doing it right now?** Three positions. The distance between them is psychological, not chronological. **TOFU — Top of Funnel** Goal: Make an unfamiliar person aware and curious. Get attention, create a follow, generate a save, build association. - They don't know the brand well yet (or at all) - The desired action is low-commitment: follow, save, share, watch again - Keywords: informational intent — "how to [problem]," "why [symptom happens]," "[topic] explained," "[problem] causes" - CTA intensity: None or Soft. A direct purchase CTA on TOFU content is mismatch. This is the "proposing marriage on a first date" error. **MOFU — Middle of Funnel** Goal: Build trust, demonstrate expertise, move someone from "aware" to "considering." Get profile visits, link clicks, DMs, email sign-ups. - They've seen content before or found it via search - Desired action is medium-commitment: click a link, join a list, engage with a question - Keywords: investigational intent — "[product/solution] review," "best [category] for [specific situation]," "[brand] vs [competitor]," "does [solution] work for [audience]" - CTA intensity: Medium. Soft ask with proof. "Link in bio" works here. "Comment [X] and I'll send you [Y]" works here. **BOFU — Bottom of Funnel** Goal: Convert. Purchase, sign up, book a call, claim the offer. - They've been in the ecosystem or are actively comparing final options - Desired action is high-commitment: buy, book, subscribe - Keywords: transactional intent — "[product name] price," "buy [product]," "[service] near me," "how to get [product]" - CTA intensity: Strong or Direct Offer. Urgency mechanics are legitimate here. Price deadlines, scarcity, social proof of the purchase moment. --- #### Formula Selection Matrix The three locked parameters intersect to determine the active formula. This is NOT a preference — it is a structural requirement of where the audience is. | Awareness Stage | Funnel | Sophistication | Active Formula | Lead Element | |---|---|---|---|---| | Unaware | TOFU | 1-3 | AIDA | Attention — pattern interrupt, curiosity | | Unaware | TOFU | 4-5 | Identity AIDA | Attention — tribe signal, belief statement | | Problem-Aware | TOFU | 1-3 | PAS | Problem recognition → agitation | | Problem-Aware | MOFU | 1-3 | PAS + soft solution | Problem → agitation → solution exists | | Problem-Aware | TOFU/MOFU | 4-5 | PAS + mechanism tease | Problem → agitation → "not the usual way" | | Solution-Aware | MOFU | 1-3 | BAB | Before → differentiated after → bridge | | Solution-Aware | MOFU | 3-5 | Mechanism-led hybrid | Unique mechanism → outcome → proof | | Product-Aware | MOFU | any | FAB + social proof | Feature → advantage → benefit + identity proof | | Product-Aware | BOFU | any | Objection-resolution hybrid | Name the objection → dissolve → offer | | Most-Aware | BOFU | any | Direct offer | Offer specifics → urgency → one-click permission | **Hybrid formula construction:** When no single formula fits cleanly, the primary formula runs the opening and core; a secondary formula provides the closing mechanics. The opening formula is always derived from the awareness stage. The closing formula is always derived from the funnel position. --- #### CTA Intensity Scale Locked from the Mode Gate. Copy construction cannot change this. | Level | When to Use | Example Mechanics | |---|---|---| | None | Unaware + TOFU | End on the hook. No ask. | | Soft | Problem-Aware + TOFU | "Save this," "Share if this hits," "Follow for [topic]" | | Medium | Solution-Aware + MOFU | "Link in bio," "Comment [X] for [Y]," "DM me [word]" | | Strong | Product-Aware + BOFU | Named urgency, social proof of purchase, limited quantity | | Direct Offer | Most-Aware + BOFU | Price + deadline + one click. No ambiguity. | --- #### Output of the Mode Gate Before leaving this phase, document the locked constraints: ``` AWARENESS STAGE: [Stage] Diagnostic evidence: [What in the inputs revealed this] What this changes: [How this stage determines the formula and what it prohibits] SOPHISTICATION LEVEL: [1-5] Diagnostic evidence: [What research revealed about market saturation] What this changes: [Whether mechanism is required, what headline type works] FUNNEL POSITION: [TOFU / MOFU / BOFU] Diagnostic evidence: [What context reveals about distance from conversion] What this changes: [CTA intensity, keyword intent tier, proof deployment] ACTIVE FORMULA: [Formula name + brief justification] CTA INTENSITY: [Level] KEYWORD INTENT TIER: [Informational / Investigational / Transactional] EMOTIONAL DRIVER: [The single leading emotion for this copy] ``` This document is internal. It feeds every subsequent phase. It is never shown to the user. --- ## Reference: platform-intelligence #### What This Phase Produces Two things: a keyword layer map that tells copy construction exactly which terms land in which copy layers for this platform, and a story architecture that defines how the scroll-stopper (artwork, thumbnail, cover) and the caption/body divide the narrative. Both outputs are hard constraints — platform-native copy is built differently per platform from the first word, not adapted from another platform's version. #### Why Platforms Are Not Interchangeable The same person in different cognitive states on different platforms needs different copy structures. Someone browsing Instagram Reels is in entertainment/discovery mode — the brain is passive, the scroll is fast, pattern interrupts work. Someone on LinkedIn is in professional validation mode — they're looking for content that confirms their expertise or makes them look informed to their network. Someone on TikTok is in community search mode — they're looking for content that matches their specific interest graph. The same message delivered with the same copy structure on all three platforms fails on at least two of them. Platform-native is not aesthetic preference — it is psychological targeting. --- #### Platform Profiles ##### Instagram (Reels, Feed, Carousel, Stories) **Cognitive Brain State:** Discovery / Entertainment mode for Reels and Explore. Relationship maintenance mode for Feed. Documentary/update mode for Stories. The scroll is fast. Attention is given in 0.3-second increments. **Primary engagement signals (what triggers reach):** - Shares: strongest signal — indicates content created genuine connection worth spreading ("I sent this to 5 people in my DMs" is algorithmic gold) - Saves: indicates high-value content worth returning to — educational, reference, or deeply resonant content - Watch-through rate on Reels: completing a Reel signals the algorithm that content held attention against all competing content in the moment - Comments: especially questions and multi-word responses — indicate opinion activation **Algorithm / SEO layer architecture (2025-26):** - Primary indexing: Caption keywords in the first 2 sentences — Instagram's AI treats captions as searchable content similar to SEO. Keywords here determine content categorization and non-follower distribution. - Secondary indexing: Alt text on static images, bio keywords for account-level topical authority - Hashtags: Reduced weight since December 2024 (following removal ceased). Use 3-5 highly relevant hashtags for search functionality — not for discovery distribution. - Text overlays on Reels: Carry limited direct indexing value but drive watch-through rate, which is a top ranking signal - Funnel position → keyword intent: - TOFU → informational keywords in caption first lines ("how to [pain]," "why [symptom]") - MOFU → investigational keywords ("[solution] that actually works," "best [category] for [specific person]") - BOFU → transactional terms + brand keywords ("get [product]," "[brand] [offer]") **Artwork / Caption story architecture:** - Scroll-stopper: Static image text overlay, Reel cover frame, or carousel slide 1 ≤5 words maximum — this is the first half of the two-part story - Caption: Picks up the story the scroll-stopper started. First 2 sentences are visible before "more" — these carry the hook AND the primary keywords simultaneously. - Two-part story pattern for Instagram: - Scroll-stopper = the QUESTION or the SHOCK or the NUMBER - Caption opening = the CONTEXT or the ANSWER or the MEANING - They must never duplicate: if the scroll-stopper says "I quit my job" the caption cannot open with "I quit my job because..." — it opens with the direct consequence or outcome that completes the story **Platform-specific hard constraints:** - Organic posts: Never include a URL in the caption body — Instagram's algorithm suppresses posts with external links. URL belongs in bio ("link in bio") or as a first comment. - Paid ads: CTA button handles the URL. Caption stays benefit/proof focused. - Stories: Ephemeral — copy must be complete without the preceding post context. Stack text + visual for accessibility (many watch without sound). --- ##### TikTok **Cognitive Brain State:** Hyper-personalized discovery mode. The For You feed is not random — it is a precise interest graph built from every interaction. Someone who sees your content is seeing it because the algorithm decided they're in a relevant interest cluster. Content feels native when it matches the tonality of that cluster — not "brand content" but "community content." **Primary engagement signals:** - Watch-through rate: finishing a video signals high value — especially on longer videos - Saves/Favorites: indicate intent to return — high-value content - Shares to DM: strongest connection signal - Comments: especially conversations in the comments section — TikTok rewards comment thread velocity - Re-watches: positive signal for short-form, very high for longer Shorts-style **Algorithm / SEO layer architecture (2025-26):** TikTok is now a search engine. Search Value is a direct ranking metric in 2026. - Spoken keywords: TikTok's algorithm "listens" via auto-captions — spoken keywords are indexed and used to match content to search queries. Keyword-rich speech in the first 15 seconds carries significant indexing weight. - On-screen text overlay: Text overlays are indexed for search relevance. Keywords in overlay + keywords spoken simultaneously create compounding relevance signals. - Caption keywords: Used for categorization and For You distribution — keyword-rich captions without keyword stuffing (natural sentence construction required) - Hashtags: Category signals — 3-5 niche-specific hashtags over generic trending ones - Funnel position → keyword intent layer: - TOFU → keywords appear in spoken intro AND overlay: "[audience] who [symptom]," "why [problem happens]," "[topic] for beginners" - MOFU → spoken + overlay + caption: "the [mechanism] that [result]," "how [product type] works for [specific situation]" - BOFU → caption + overlay + CTA: "[product] review," "where to get [product]," "[offer] expires [when]" **Artwork / Caption story architecture:** - Scroll-stopper: Frame 1 text overlay — appears on static preview in For You feed before autoplay begins. ≤4 words maximum. Upper third of frame. - Hook (spoken + overlay): Words said AND displayed in the first 2-3 seconds. These must be identical or complementary — not redundant. The spoken hook and the overlay hook are two separate attention vectors pointed at the same entry emotion. - Caption: Continues the narrative from the video — treated as a search object, not a social caption. Natural sentences with embedded keywords. - Two-part story for TikTok: - Frame 1 overlay = the PATTERN INTERRUPT or IDENTITY SIGNAL - Spoken hook (first sentence) = the SPECIFIC CLAIM or PROMISE THAT COMPLETES IT - Example: Overlay says "For people who quit" / Spoken opens with "If you've started this [exactly three times] and stopped, what I'm about to say will feel uncomfortably specific." **Platform-specific hard constraints:** - Copy must feel community-native, not brand-broadcast. "Brand voice" on TikTok means the person who made this video is a real person, not a marketing department. - The platform deprioritizes content with third-party watermarks (TikTok Creative Center is neutral; other platform watermarks are penalized) - Organic: No URL in caption. Link in bio or TikTok Shop integration. - Sound: Even with text overlay, spoken words carry indexing weight. Silent videos lose a significant indexing layer. If text-only, add even minimal voice-over. --- ##### LinkedIn **Cognitive Brain State:** Professional validation / reputation management mode. Users are curating their professional identity with every interaction. Content that makes them look informed to their network gets shared. Content that feels like it was written for their peers (not for them) gets scrolled. Personal storytelling outperforms broadcast authority. **Primary engagement signals:** - Comments: especially thoughtful responses — LinkedIn's algorithm heavily weights comment quality and velocity in the first 60 minutes ("golden hour") - Shares: indicates content is worth professional identity association — someone shared it to be seen as the person who surfaces good content - Dwell time: LinkedIn measures how long you read, not just scroll past - Saves: private but tracked — signals high-value professional reference content **Algorithm / SEO layer architecture (2025-26):** - Caption keywords: LinkedIn treats posts as professional search content. Niche-specific terms, role titles, industry terms carry indexing weight. - No hashtag distribution (LinkedIn removed hashtag following in 2024 for most accounts). Hashtags still function as search signals — 3 maximum, niche-specific. - Topic consistency: LinkedIn's algorithm builds a "topic authority" score per creator. Consistent posting in one professional domain outperforms generalist content. - Funnel position → keyword placement: - TOFU → Professional vocabulary in first 3 lines, topic-specific keywords, visible before "see more" cutoff - MOFU → Specific insight keywords, niche problem vocabulary, decision-maker language - BOFU → Service/product terms + specific outcomes + direct attribution **Artwork / Caption story architecture:** LinkedIn is caption-first. The image or video is secondary. The first 3 lines of the caption are the scroll-stopper — they appear before the "see more" cutoff. - "Scroll-stopper" on LinkedIn = first 2-3 lines of text (approximately 200 characters) - These lines function identically to the artwork on other platforms — they must stop the scroll and create a reason to click "see more" - The two-part story structure on LinkedIn: - First 2 lines = the HOOK (provocative claim, specific insight, counterintuitive statement) - Everything after "see more" = the PROOF, NARRATIVE, and CTA - The break is the story: everything before "see more" promises something. Everything after delivers it. **Platform-specific hard constraints:** - Never include a URL in the caption body — LinkedIn's algorithm significantly suppresses posts with external links. URL belongs in the first comment, not the post itself. - Copy should sound like a person with expertise talking to their peers — not like an ad and not like a brand broadcast. The phrase "I" outperforms "we" consistently. - Professional storytelling structure (I did [thing] → this happened → here's what I learned / here's what this means for you) consistently outperforms direct promotion. --- ##### Facebook (Feed, Groups, Ads) **Cognitive Brain State:** Social scroll / community check-in mode for organic Feed. Passive media consumption for Groups. High-intent discovery for Ads (Meta's targeting is precise enough that BOFU Facebook ads reach near-product-aware audiences). **Algorithm / SEO signals:** - Groups: highest organic reach on Facebook — community posts in Groups outperform Page posts by significant margin for organic distribution - Shares: most powerful distribution signal on Facebook — indicates content crossed the "worth putting on my wall" threshold - Video watch time: especially native video (not YouTube links) - Comment debates: Facebook surfaces content that generates back-and-forth comment threads **Artwork / Caption story architecture:** - Facebook is text-heavier than Instagram — longer captions are read more often on Facebook than on Instagram - The hook (first 1-3 sentences before "See more") must do the same work as Instagram's scroll-stopper — but in text form - Problem-forward openings historically perform: "If you've ever [specific frustration]..." works because it creates immediate pattern recognition - Paid ads: Facebook/Meta ads use the Primary Text field (caption), Headline, and Description. Primary Text carries the story. Headline carries the mechanism or outcome. Description adds the proof element. These are three separate copy jobs, not one caption split. **Platform-specific hard constraints:** - Algorithm penalizes "engagement bait" — explicit asks for likes/shares ("like this if...") trigger distribution suppression - Video auto-plays without sound in Feed — copy that works without audio via text overlay is required for video formats --- ##### X / Twitter **Cognitive Brain State:** Real-time news / opinion mode. Moving fast. High cynicism threshold. Credibility is carried by the account, not the post. New accounts without established presence have limited organic reach. **Algorithm signals:** - Replies and quote posts: Twitter/X heavily weights replies (conversation) and quote posts (commentary by others — positive or negative) - Bookmarks: equivalent to saves — high-value content signal - Engagements from high-follower accounts: disproportionate distribution boost **Artwork / Caption story architecture:** - Twitter is text-first. 280 characters is the scroll-stopper, the hook, and often the entire message simultaneously. - Thread first post = the headline + the teaser ("here's what I found out 🧵") - Thread is the article — each tweet advances the argument or story - Single-tweet copy must be complete without any click required for TOFU content - BOFU single-tweet copy: claim + proof element + CTA in 280 characters --- #### GEO Layer Summary Table | Platform | Layer 1 (Primary Index) | Layer 2 | Layer 3 | CTA Placement | |---|---|---|---|---| | Instagram | Caption first 2 sentences | Alt text / Bio | Overlay (engagement) | Bio link / First comment | | TikTok | Spoken audio (first 15s) | Text overlay | Caption | Bio link | | LinkedIn | First 3 caption lines | Caption body | Hashtags (3 max) | First comment | | Facebook | Headline (paid) / First lines (organic) | Primary text / Caption | Description (paid) | CTA button / Comment | | X/Twitter | Tweet text | Thread opening | Image alt text | Linked tweet in thread | #### Two-Part Story Architecture Quick Reference | Platform | Scroll-Stopper | Caption/Body Role | |---|---|---| | Instagram Reels | Cover frame text (≤5 words) | Completes the story. Keywords in first 2 lines. | | TikTok | Frame 1 overlay (≤4 words) | Spoken hook + caption continue the story. | | LinkedIn | First 2-3 caption lines (no image needed) | Proof + narrative after "see more" | | Facebook Feed | Image headline OR first 1-3 caption lines | Story body — longer is acceptable | | Twitter/X | The tweet itself | Thread unfolds the argument | --- ## Reference: segment-intelligence #### What This Phase Produces A specific, evidence-grounded profile of the exact human this copy is written for — not a demographic label, not a persona template, but a living picture of what this person currently believes, fears, has tried, blames, and wants to feel about themselves. Every constraint this profile produces is downstream copy material. The profile is not a document about the audience — it is a set of constraints on what copy is allowed to say. #### Why Depth Matters Surface-level profiling produces surface-level copy. Knowing that the audience is "25-35 year old women interested in fitness" does not tell you whether to lead with shame-activation or aspiration, whether they've been burned by the exact promise you're about to make, whether "coach" language triggers trust or skepticism, or whether calling them "busy moms" activates belonging or condescension. The 7-layer framework below exists because every layer answers a different question that copy construction needs answered. --- #### Step 1: Sharpen the Segment The label the user provides — "founders," "moms," "fitness beginners," "small business owners" — is a starting point, not the target. ##### 1.1 The Specificity Interrogation Start with the label and ask: - **Who within this segment** is most likely to encounter this content and be moved to act? Not "entrepreneurs" — "first-time solo founders at the 6-month mark who've been doing everything themselves and are starting to feel the ceiling." - **At what stage of their journey** are they? Pre-problem? Mid-struggle? Post-failure? About to make a purchase decision? - **What specific triggering event** brought them to this moment? A failed attempt? A milestone not reached? A friend's success? Something they read or saw? - **What specific version of the problem** do they have? The general category is not the answer. "I want to lose weight" is not the same as "I've tried three diets and I'm starting to believe this is who I am." Search for: "[segment] [problem category] experiences," "[segment] [category] Reddit," "[segment] [product] reviews," "[segment] [topic] community forums." Build the sharpened decision-maker — one specific person — and record why the sharpening changes messaging decisions downstream. ``` SHARPENED SEGMENT: Starting label: [e.g., "fitness beginners"] Sharpened to: [e.g., "Women 30-45 who've started and quit fitness programs at least once, currently in a 'I'll try again differently this time' mental frame"] Triggering event: [What brought them here now] What this changes for copy: [Specifically what this sharpening means for how we write] ``` --- #### Step 2: The 7-Layer Psychographic Profile Build every layer from fresh research — not from assumptions about what this person "probably" thinks. Every layer requires evidence from actual sources. ##### Layer 1: Professional / Social Identity **Question:** How does this person describe themselves — and what community do they belong to? Search for: Their LinkedIn language if professional, their community spaces (which Subreddits, which Facebook groups, which Slack communities), what they say when introducing themselves. Capture: - How they describe their role or situation in one sentence - The community they affiliate with (not just demographically — culturally) - What they're proud of professionally or personally - Whether they identify as inside or outside the mainstream of their category (e.g., "I'm not like most people who do [thing]") **What this changes for copy:** Identity language in copy either activates belonging ("yes, this is for me") or triggers rejection ("this isn't for someone like me"). Copy that uses the wrong identity label loses the reader in line 2. ##### Layer 2: Motivation Architecture **Question:** What is this person trying to achieve — and what does success look like specifically for them? Search for: What outcomes this segment talks about when they describe winning, what KPIs or milestones they measure themselves against, what they tell others when things are going well. Capture: - The specific outcome they want (not the category outcome — the personal version) - What success looks like to them in concrete terms — what changes, what disappears, what they can finally say or do - Whether their motivation is approach-oriented (I want to gain X) or avoidance-oriented (I want to stop feeling Y) — this determines the dominant emotional driver - What "good enough" looks like vs. what "transformed" looks like **What this changes for copy:** Approach-motivated audiences respond to outcome copy ("here's what you get"). Avoidance-motivated audiences respond to relief copy ("here's what you stop experiencing"). Getting this wrong means the right result lands wrong. ##### Layer 3: Current Emotional Baseline **Question:** What is this person feeling about this topic, problem, or category RIGHT NOW — not in general, not historically, but at this moment in time? Search for: Recent Reddit threads, comment sections, community posts, Twitter/X conversations from the past 30-90 days on this topic. What is the dominant emotional register? Capture: - The dominant emotion: exhaustion, skepticism, anxiety, cautious hope, excitement, shame, frustration, or something more specific - Whether that emotion is getting louder or quieter (trending toward action or retreating) - What specific recent event or trend is shaping their current emotional state (a platform change, a cultural moment, a news cycle, a pattern they've been seeing) - Whether they feel alone in this or surrounded by others feeling the same way **What this changes for copy:** The emotional baseline determines the entry emotion in Phase 3 synthesis. Copy that leads with the wrong emotion creates dissonance — the reader feels misunderstood and stops reading. If they're exhausted, opening with high-energy inspiration feels tone-deaf. If they're cautiously hopeful, opening with deep problem agitation reverses the goodwill they just extended. ##### Layer 4: Fear & Pain Mapping **Question:** What is this person afraid of — both as a result of acting AND as a result of not acting? Search for: "[segment] afraid of," "[segment] worried about [category]," "[segment] scared," Reddit threads where they describe hesitation, "what's stopping you" community posts. Capture the four-quadrant fear map: ``` ACTING WRONG FEAR: [Specific to this person — e.g., "Buying another course that doesn't apply to my situation, spending $300, and still being stuck in 6 months"] NOT ACTING FEAR: [e.g., "Everyone else figures this out and I fall further behind while I was 'being careful'"] CURRENT PAIN: [What's actively hurting them right now — specific, named frustration] PAST PAIN: [What they've already tried that failed — and what that failure taught them about themselves] ``` **What this changes for copy:** The fear architecture determines whether copy leads with "here's what you risk by not acting" (loss-framing for avoidance motivators) or "here's what you get by acting" (gain-framing for approach motivators). The past pain layer is especially important: if they've been burned by the same promise before, leading with that promise repeats the burn. Copy must acknowledge the past attempt before making the current pitch. ##### Layer 5: Blocking Beliefs **Question:** What does this person believe about themselves — not about the product, not about the category, but about THEMSELVES — that makes the action feel unnecessary, premature, or impossible? This is the deepest layer and the most overlooked. It is not about market skepticism. It is about self-concept. Search for: "[segment] [result] Reddit 'not for me,'" "[segment] 'tried this,'" "[segment] 'can't,' 'won't work for me,' 'different situation,'" phrases starting with "I'm the kind of person who..." Common blocking belief patterns: - **Uniqueness belief:** "My situation is different from everyone this works for" → Copy must address the specific way they believe they're different - **Worthiness belief:** "I'm not the type of person who succeeds at this" → Copy must show people with exactly their self-concept having the outcome - **Timing belief:** "This isn't the right time for me to start this" → Copy must address what keeps the timing from ever being right - **Effort belief:** "The results are real, but I won't be consistent enough" → Copy must address the effort requirement directly, not promise effortlessness - **Burned belief:** "I've tried something like this and it didn't work" → Copy must acknowledge the past failure and differentiate the mechanism ``` PRIMARY BLOCKING BELIEF: [One sentence — what they believe about themselves] Evidence from research: [Where this showed up] What this changes for copy: [Whether to lead with mechanism differentiation, identity confirmation, or acknowledgment of past failure] ``` **What this changes for copy:** The blocking belief is the invisible wall between the audience and the action. Copy that doesn't address it makes a compelling pitch to someone who has a reason not to believe it. The blocking belief doesn't need to be named explicitly — but the copy must dissolve it structurally, either through mechanism differentiation, identity mirroring, or acknowledgment of the specific way this is different from what failed before. ##### Layer 6: Identity Stakes **Question:** What identity is threatened or confirmed by taking this action — and which matters more to this person right now? Every purchasing or action decision carries identity weight. "If I buy this, what does that say about who I am? If I don't, what does that say?" Search for: How this segment talks about the product category in terms of who uses it, what it signals about a person, whether there are negative associations (e.g., "only people who can't do it themselves buy a program"), positive associations (e.g., "serious people invest in their growth"). Capture: - **Positive identity signal:** What does taking this action confirm about them? ("I'm someone who takes their [X] seriously," "I'm part of the group that [does/has/is]") - **Negative identity threat:** What does taking this action risk? ("It means I need help," "It means I've failed at doing it alone," "It means I'm one of those people who [category]") - **Social identity dimension:** What does this action signal to others they care about? ("People will see me as," "My [peer group] will think") **What this changes for copy:** At Sophistication Level 5, identity IS the message. At any level, copy must work with — not against — the audience's identity stakes. If taking the action carries a negative identity threat (admitting you need help, spending money on something basic, trying something others might mock), copy must either resolve the threat ("people who [positive identity marker] choose this") or reframe the action as consistent with who they already are. ##### Layer 7: Language & Discourse Patterns **Question:** What exact words, phrases, and constructions does this person use when they're sitting with this problem — not when they're writing professionally, but when they're talking to themselves or their community? This is the most mechanical layer and the most important for word-level copy decisions. Search for: Comment sections on competitor content, Reddit threads, Facebook group posts, product reviews (especially 3-4 star reviews — they explain both what worked and what didn't), Twitter/X conversations. Mine 30-50 comments per source, not just the top 10. Capture: - Minimum 10 real phrases, verbatim, from real people (not paraphrased into marketing language) - Identify the vocabulary register: casual/conversational, technical/insider, formal, emotional - Identify exhausted words in the category — terms so overused they've lost activation value ("game-changer," "transform," "revolutionary" for some categories) - Identify in-group signal words — the specific terms that mark someone as inside the community - Identify the emotional words they use most: do they describe the problem in terms of feeling (frustrated, stuck, overwhelmed) or outcomes (behind, losing, failing)? ``` CAPTURED PHRASES (minimum 10): 1. "[Exact quote from real source]" — [Source type] 2. "[Exact quote]" — [Source type] ... (continue to 10+) VOCABULARY REGISTER: [Casual / Technical / Emotional / Mixed] EXHAUSTED TERMS TO AVOID: [List] IN-GROUP SIGNAL WORDS: [List] DOMINANT EMOTIONAL VOCABULARY: [Feeling words or outcome words?] ``` **What this changes for copy:** Copy that uses the audience's actual words converts because it signals "this person is in my world" — copy that paraphrases into marketing language signals "this is an ad." Every word in the final copy should either come directly from the captured phrases or be derived from their register and vocabulary. If a word in the final copy doesn't trace back to the language inventory, it earns scrutiny. --- #### Segment Synthesis Output After all 7 layers are complete, produce the synthesis constraint document: ``` SEGMENT SYNTHESIS: Sharpened target: [One specific person — not a demographic, a human] Blocking belief: [One sentence — what they believe about themselves that blocks action] Identity stakes: [What taking action confirms / threatens for them] Current emotional baseline: [Dominant emotion about this topic right now] Dominant driver: [Approach (gain) or avoidance (loss)] Past pain note: [Have they been burned by similar promises? What failed?] Language inventory: [The 10+ captured phrases] Register: [The vocabulary register for all copy] Exhausted terms: [Do not use these] What this means for Phase 3: → Entry emotion: [What copy must make them feel in lines 1-2] → Mechanism note: [Must copy differentiate the mechanism from past failed attempts?] → Identity frame: [Must copy confirm identity belonging or resolve identity threat?] → Proof type: [Which evidence format this audience trusts] ``` --- ## Reference: verification-attacks #### What This Phase Produces Evidence that the copy works — not an opinion that it works. The difference is structural: evaluative checks ask "is this good?" and the model that produced the copy answers "yes." Generative attacks require producing proof of quality that is independently verifiable. A rubber stamp costs nothing. Generating evidence costs real cognitive work — and if the evidence can't be generated, the copy fails. #### The Standard The audit should catch at least ONE issue. If it catches nothing, run more aggressively. Copy that survives every attack below has earned its production status. --- #### Attack 1 — Transplant Test **What it requires:** Take the scroll-stopper and the hook (first two lines of the caption). Try to run them verbatim for a direct competitor's product targeting the same audience. **How to run it:** 1. Name a specific competitor in this category 2. Insert their brand/product name into the scroll-stopper and hook 3. Read it as if it appeared on their account **Pass/Fail:** - If the transplanted version would work for the competitor → **FAIL. The copy is generic.** The scroll-stopper or hook is relying on category language, not brand-specific or mechanism-specific language. Go back to Phase 3 (Positioning Synthesis) and sharpen the mechanism differentiation. - If the transplant breaks — if the specific language, mechanism reference, or identity framing only makes sense for this brand/product — → **PASS.** **Why this matters:** Generic copy converts for everyone, meaning it converts for no one specifically. If a competitor could run this exact copy and it would work, you've done their marketing for them while also doing yours. The only copy worth producing is copy that belongs to this brand. --- #### Attack 2 — Formula Element Naming **What it requires:** Name the specific formula element each section of the copy is executing. **How to run it:** For every paragraph or structural section of the copy, name: - Which formula is running (PAS, AIDA, BAB, FAB, Hybrid) - Which specific element of that formula this section executes (Problem / Agitate / Solution; or Before / After / Bridge; etc.) - What constraint from the constraint set this section satisfies **Pass/Fail:** - If you can name the formula element for every section → **PASS.** - If a section doesn't correspond to any formula element → **FAIL. Remove or rebuild it.** Sections that "feel right" without a formula function are filler — they dilute the structural integrity of the copy without adding persuasive force. - If the formula element sequence doesn't match the locked formula → **FAIL. Restructure.** A PAS that goes Problem → Solution → Agitate has lost the mechanic that makes PAS work. **Why this matters:** Formula violations are often invisible until this test. Copy that feels good on a read-through may have reorganized its elements in a way that breaks the psychological sequence. The formula is not arbitrary structure — each element exists because of what the previous element set up. --- #### Attack 3 — Language Inventory Trace **What it requires:** For every substantive word choice in the copy, either trace it to the captured language inventory (Layer 7) or justify its inclusion. **How to run it:** 1. Read the copy word by word 2. For every word that is NOT a function word (a, the, is, and, etc.) — ask: Is this word in the captured language inventory? If not, is it derived from the vocabulary register of the inventory? 3. Identify any words that come from marketing defaults rather than the segment's voice (typical marketing defaults: "game-changing," "revolutionary," "powerful," "amazing," "transform," "unlock," "skyrocket," "journey") **Pass/Fail:** - If all substantive words trace to the inventory or the vocabulary register → **PASS.** - If 3+ words are marketing defaults that don't appear in the inventory → **FAIL.** These are the words that make copy feel like it was written "at" the audience rather than for them. Replace with inventory language. - If the vocabulary register is inconsistent (some lines sound like the segment, others sound like a brand) → **FAIL. Rewrite for register consistency.** **Why this matters:** Audiences recognize their own language immediately. Copy that uses their words feels like it comes from inside their world. Copy that uses marketing language feels like an ad. The language inventory isn't a stylistic preference — it is the mechanism by which copy either passes or fails the "this person gets me" test in the first three seconds. --- #### Attack 4 — Two-Part Story Test **What it requires:** Read the scroll-stopper alone. Then read the caption/hook opening alone. Test whether they form a complete two-part story or whether they duplicate, contradict, or fail to connect. **How to run it:** 1. Read the scroll-stopper aloud. Stop. Ask: Does this raise a question, create a gap, or leave something unresolved? 2. Read the caption's first two lines. Ask: Do they resolve, extend, or complete what the scroll-stopper raised? 3. Check for duplication: does the caption repeat information from the scroll-stopper? **Pass/Fail:** - If the scroll-stopper raises something and the caption opening resolves it → **PASS.** - If reading the scroll-stopper alone is satisfying — complete thought, no open loop → **FAIL.** A complete scroll-stopper removes the reason to read the caption. Rebuild with an open loop. - If the caption opening repeats the scroll-stopper's information → **FAIL.** Example: Scroll-stopper says "3 years to figure this out" / Caption opens "It took me 3 years to figure this out" — duplication collapses the two-part system into one. Rewrite so each part carries new information. - If the scroll-stopper and caption feel disconnected — like they were written separately → **FAIL.** They should feel like the first and second sentence of the same thought. **Why this matters:** The two-part story is the mechanism that earns attention after the scroll stops. The scroll-stopper stops the thumb. The caption opening closes the gap the scroll-stopper opened. If the gap isn't there, there's nothing to close. If the gap is already closed, there's no reason to stay. --- #### Attack 5 — Awareness Stage Alignment **What it requires:** Read the copy as a member of the diagnosed awareness stage. Ask: does the copy require knowledge this person doesn't have yet, or does it try to sell something this person isn't ready to buy? **How to run it:** For the diagnosed awareness stage, test the alignment: **If Unaware:** Does the copy mention the product or assume knowledge of the problem? - A good Unaware-stage copy could be read without knowing what's being sold - If "buy [product]" appears anywhere → misaligned. Remove or rebuild. **If Problem-Aware:** Does the copy skip the agitation and jump to the solution? - If the copy moves from naming the problem directly to "here's the answer [product]" without agitation → too fast. Add agitation. - Does the copy assume they're already searching for solutions? If yes → too fast. **If Solution-Aware:** Does the copy explain the problem as if they don't already know it? - Agitating a problem to a Solution-Aware audience is condescending — they've already acknowledged the problem. Skip straight to mechanism differentiation. **If Product-Aware:** Does the copy try to build awareness of the problem or category? - Product-Aware audiences have already researched. Copy that explains the category insults their intelligence. Go directly to "why this product for this specific situation." **If Most-Aware:** Does the copy try to educate or persuade rather than activate? - Most-Aware audiences don't need to be convinced. They need the last reason to say yes. Copy that still agitates the problem is wasting their time. **Pass/Fail:** - If the copy matches the diagnosed awareness stage at every structural level → **PASS.** - If any section assumes knowledge the diagnosed stage doesn't have → **FAIL.** Remove that section. - If the CTA asks for an action the diagnosed stage isn't ready for → **FAIL.** Reduce CTA intensity. --- #### Correction Protocol When the user gives a correction after seeing output: **Step 1: Diagnose the phase.** Which phase produced what the user is correcting? - Copy feels generic → failure in Segment Intelligence (thin language inventory) or Positioning Synthesis (mechanism not differentiated) - Copy is misaligned with the audience → failure in Mode Gate (wrong awareness stage diagnosed) or Segment Intelligence (wrong blocking belief or emotional baseline) - Copy sounds like an ad → failure in language inventory (marketing defaults, not segment language) - Hook doesn't land → failure in hook construction (wrong emotional driver for entry state) - CTA is wrong strength → Mode Gate needs updating (funnel position was wrong) **Step 2: Fix immediately.** Rebuild the specific element from the phase that failed. Don't adjust the wording. Don't polish. Rebuild from the constraint. **Step 3: One sentence explanation.** What changed and what phase produced the fix. **Never ask permission to fix what the user already identified as wrong.** The correction is the instruction. Execute it.