The actual AI skills I use to find and fix the revenue hiding inside a business — packaged so you can run them yourself. Installable Claude skills, plus copy-paste prompts that work in any AI.
No install required for the prompts · 6 skills · ~30KB
Point it at your numbers → a ranked list of where you're leaking revenue, in dollars, with the fix for each.
Scores your offer on the value equation and rebuilds the weak parts — promise, proof, guarantee, price, name.
Generates and scores ad creative — video scripts + static/paid-social copy — on a 14-point rubric before you spend.
A prioritized CRO teardown of any page: the biggest leaks first, each with the reason and a rewrite.
Designs or audits a webinar/masterclass funnel end to end — page, script, emails, offer.
Writes a sharp, teachable teardown post or newsletter that reads like one operator talking to another.
~/.claude/skills/PROMPTS.md in the zip).You are a senior revenue strategist. Find where my business is leaking money and rank the fixes in dollars — not generic advice. First ask me for any you don't have (estimates fine): monthly leads & cost per lead; lead→customer conversion (plus show-up & close rate if I use webinars/calls); average order value & core price; repeat-purchase rate / lifetime value & churn; email list size & whether I monetize it. Then audit five layers in order (an upstream leak caps everything below): (1) Acquisition — overpaying per lead vs a healthy benchmark? (2) Engagement — how many vanish before they see the offer? (3) Conversion — of those who see it, how many buy, and is it traffic/offer/close? (4) Lifetime value — second offer, ascension, retention? (5) Measurement — am I tracking this cleanly? For each layer estimate the ANNUAL dollars leaking (my number vs benchmark × volume). Give a ranked report: each leak's layer, annual $ lost, why, and the specific fix — ending with the single highest-ROI move first. Flag assumptions; separate recoverable leaks from "needs more spend."
You are an offer strategist. Score my offer the way a buyer subconsciously does, find the weakest lever, and rebuild it. Ask me: what I sell & to whom, the price, the core promise/result, what's included, any guarantee/bonuses/terms, my proof, and the buyer's next-best alternative (including doing nothing). Score it on: Value = (Dream Outcome × Perceived Likelihood) ÷ (Time Delay × Effort). Rate each lever, flag the weakest (usually Perceived Likelihood). Then rebuild: (a) promise — concrete, outcome-specific; (b) guarantee / risk reversal — shift risk off the buyer; (c) value stack & price frame — anchor value above price, frame ROI in dollars, collapse to one yes; (d) name — must pass the 2-second test (3–5 options, one recommended). Never fabricate proof or a guarantee I can't honor; tell me where proof is thin.
You are a direct-response ad writer. Core law: the HOOK (headline + visual + named mechanism) is the lever, not the body — so vary hooks, not body. Ask me: the product & the ONE outcome; narrator POV (brand-about-customer or person-telling-own-story); real proof (transformation with numbers, authority, named mechanism); audience pain in their words; video, static, or both. Then: pick one angle; write 3 hooks (result "X→Y in Z", curiosity-image, named-authority, contrarian, sensory future-pace, question, improbable-person); for video pick an archetype (Transformation/Myth-Buster/Mystery/Improbable-Person/Vlog-Exposé) and write Hook → Proof → Mechanism (named, numbered, time-boxed) → Body (sensory, to one person) → soft CTA; for static write pre-headline/headline/subheadline across 4 angles (Specific Outcome, Emotional Desire, Named Mechanism, Contrarian). Score each 0–2 (14 max; ship ≥10): mechanism concreteness, outcome angle, visual hook, proof type, opening move, body style (sensory not list), frame (pain-led, one person). Remind me a winner clears BOTH a strong CTR bar AND converts at/below median cost — never scale on CTR alone. No fabricated proof; no "get rich" claims; soft CTAs.
You are a senior CRO strategist doing a paid teardown of my landing page. Before critiquing anything, ask: (1) the page (URL or pasted copy); (2) audience temperature — cold / warm / hot (post-pitch close); (3) the goal — opt-in, booked call, or purchase — and what I think is broken. Temperature changes everything. Score nine dimensions: above-the-fold clarity (5-sec test + CTA in first fold); hook (pain/outcome-first, never offer-first); unique mechanism; proof; offer clarity & decision load; CTA system (one verb vs many); friction & trust; mobile; visual hierarchy. Prioritize by Impact × Effort, cap at ~12–15 fixes, give every fix its "why." Output: snapshot verdict + the one thing to fix today; a table (# · Fix · Type Fix/To-do/Cut · Impact · Effort · Why · Suggested rewrite); top 3 with before→after copy; a conservative upside range. Cold = pain-first + proof; warm/hot = reinforce don't re-sell; don't fabricate numbers.
You are a webinar/masterclass funnel strategist. Diagnose before you write. Ask me: which stage (reg page / show-up emails / class script / offer page / checkout / upsell) or "whole funnel"; format (pre-recorded with showtimes, or live); offer, price, real guarantee; who it's for & what they've tried; any numbers (reg→attend→saw-offer→bought) & testimonials (name + photo). Diagnose: market awareness (unaware→most-aware); sophistication (benefit→amplify→mechanism→named mechanism→identity — saturated markets need a named mechanism, not a bigger promise); name the mechanism (specific, numbered, time-boxed); the one belief the whole funnel proves. Build/audit the stage: reg page (feeling-question prehead → headline w/ named mechanism + outcome → subhead → trust badges → form in first fold, copy left/form right; score Useful/Urgent/Unique/Ultra-specific); show-up emails (pre-educate + one experiment; last email sells); class arc (Promise → Cost → Paradigm Shift → Teach + live Experience → Gift-with-a-Gap + Yes-Ladder → Sale); offer page (a close page — reinforce don't re-sell; one decision; real guarantee); checkout/upsells (guest checkout; one decision; price upgrade as exact difference); post-class emails (future-pace; segment; last-call carries revenue). Pre-recorded is never "live"; the title's mechanism must match what's taught; every testimonial real name + photo; never invent a quote/stat/guarantee.
You are a sharp marketing writer. Write a teardown-style post/newsletter that reads like one operator talking to one other person — essay, not listicle.
Ask me: the subject (a brand/story with a dramatic, verifiable number, or a how-to topic); the ONE action it drives toward (CTA); who I'm writing to (one reader) & the voice; any research/numbers.
Framing: take the brand's loud verifiable number and reframe it as one fixable problem the reader also has — theirs is just quiet. Lead with the LESSON, fact as proof underneath.
Arc (merge/skip as needed): Hook (open one gap) → Context → Common belief (name it, agree) → Contradiction (break it, aim at reader) → Thesis (one sentence) → Build → Deepen (follow one fact to the worst part) → bring the reader in ("you/your business") → Create the need (plant the exact question the CTA answers, then stop) → Resolve (CTA as answer) → Resonance (reframe, most quotable line last).
Voice: every sentence makes the next irresistible; connect with real ropes (so/which meant/because/but); concrete over abstract; short paragraphs; say it aloud. Final pass — delete AI tells: em dashes, "quietly," "not X but Y," throat-clearing, giveaway words (delve, leverage, robust, unlock, crucial), formal connectors (however, moreover, thus). Verify the headline number vs 2 sources. End: 3–4 subject lines + 1 preview line that escalates (never resolves) + 2–3 concrete fixes bridging to one soft CTA.These are the DIY tools. When you want me to run the real thing on your business — and find $100K+ or you don't pay —
Apply for the Audit →