Jonathon's AI Wiki

Lead Magnet Creation with Claude Code (Brandon Storey course)

17 min read

Source: raw/Free_Course_-_How_to_Make_Lead_Magnets_with_AI_Claude_Code_Tutorial.md Creator: Brandon Storey (“Copywriting Coach” YouTube channel) URL: https://youtu.be/T5x-o9bBKdg Published: 2026-04-15 Duration: 1:55:54 Platform: YouTube

A nearly two-hour walk-through of an end-to-end lead-magnet pipeline by Brandon Storey, a self-described 7-year six-figure copywriter who runs the paid Six Figure Copy Academy on school.com. The video is filed in ai-marketing because it’s a marketing playbook, not a Claude Code tutorial — Claude Code is one of seven tools used. Distinctive value: it shows the whole funnel, not just the prompt. Six sequential phases (ideation → PDF production → triple lead magnet → email automation → landing pages → social distribution) glued together with Zapier and ManyChat, with conversion CTAs added at every stage so the lead magnet doubles as a paid-offer pitch from the moment of download.

Key Takeaways

  • A lead magnet is “something you give to people in order to turn them into a lead” — free guide, checklist, PDF, ebook, video, discount code, or AI tool. The job is list growth + nurture + a chance to convert.
  • Practitioner numbers (self-reported from the speaker’s own Beehiiv): 1,400 subs from a swipe-file lead magnet, 2,700 from “How I landed my first client in 45 days,” ~500 from a free welcome-email template. Single lead magnets can carry thousands of subs each.
  • The recommended ideation move: dump existing customer/audience signal (private community questions, coaching-call transcripts via Fathom/equivalent, course outlines) into Claude and ask for angles and ideas, not finished lead magnets. Then human picks the angle. Then chain-prompt to “now build idea N.”
  • Source-of-truth for ideas often turns out to be the creator’s own paid course — break a piece off, give it away. The video’s worked example takes a “12 proven headline templates” cheat sheet that already lived in Brandon’s paid academy and turns it into a free PDF + custom GPT.
  • Production phase: Claude Code converts a plain Google Doc into a branded PDF in one shot. Speaker layers a downloaded “design skill” (calls it “a buff” — graphic-designer rules from a public source) plus a brand-guidelines file Claude Code locates on disk by name when asked. First-pass output gets to ~70-80%; conversational tweaks (“make it more centered,” “wider margin,” “transparent background”) close the gap.
  • Same content ships in three formats from one source — Google Doc + PDF + custom ChatGPT GPT. Speaker calls this a “triple lead magnet.” Trade-off: more surface area for distribution at the cost of one-time setup ~30 min for the GPT.
  • Every conversion surface gets a paid-offer CTA: end of the lead magnet PDF, the delivery email body, the landing page form, the thank-you page, and the ManyChat DM follow-up. The PDF’s CTA copy is generated by Claude Code from Brandon’s school.com about page (i.e., AI rewrites his existing sales page into a button + bullet block).
  • Tech stack he runs is multi-tool, not all-in-one: Beehiiv (email), Kajabi (landing/thank-you pages), Zapier (Kajabi form → Beehiiv API trigger), Vizard.ai (caption short-form video), ManyChat (Instagram comment-to-DM automation). All-in-one alternative he names: Go HighLevel.
  • The “AI copywriter as data synthesizer” framing: three sub-skills — (1) extract data (private-community questions are juiciest, then non-buyer audience comments, then competitor reviews/Reddit), (2) push data to Claude with the right prompt, (3) refine output with feedback loops. Most “AI doesn’t work” complaints fail at step 1.
  • Editorial caveat — the video closes with a hard pitch for the speaker’s $X/mo Six Figure Copy Academy (“first paid client in 90 days or 100% money back”). Standard lead-magnet-funnel discipline applies: the video itself is a lead magnet for the academy.

The Six-Phase Workflow

Phase 1 — Ideation (Claude on the web)

Don’t start by writing. Start by listing what your audience already asks for, in their own words. Brandon’s hierarchy of market-research data, from juiciest to weakest:

  1. Questions/posts inside your paid community (highest signal — these are people who already trust you with money)
  2. Comments and emails from non-buyers in your audience
  3. Competitor comments / reviews / Reddit threads in your space

Then paste batches of that signal into Claude (web, Opus 4.6 in the demo) with this kind of ask: “Give me 10 angles and ideas for lead magnets for my audience. Audience is X, here are some questions they’re asking…” + paste. He uses Fathom to grab coaching-call transcripts and pastes those in too — “I could have spoken this out, but I’ll just paste.”

Tool note: he speaks the prompt rather than typing — uses WhisperFlow for voice-to-text. Argues this is the easiest way to use AI: “stop typing to AI, start speaking to AI” — you can ramble and it still understands.

Output of Phase 1 is a list of angles, not a finished lead magnet. Then human taste picks one. Then chain-prompt: “Great, let’s go with idea N. Here’s more data. Make it.”

Phase 2 — Production (Claude Code)

The worked example: a “12 proven headline templates” cheat sheet that already lived as a plain Google Doc in his paid academy. Goal — turn it into a branded PDF for free download.

Claude Code workflow:

  1. Download the Google Doc as PDF (Claude Code wants PDF/PNG/JPEG/GIF, not DOCX).
  2. Drop the PDF into Claude Code with this prompt shape: “Create a pretty PDF version of this headline-template checklist called Juicy Hooks. This is for freelance copywriters to learn how to write better headlines so they can land more clients. Make it a prettied-up version of my plain-text Google Doc.”
  3. First pass produces a generic-styled HTML-rendered-as-PDF. Open it, look at it, talk to it: “I can’t open it, can you fix that?” → it fixes. “Looks good but not on-brand.”
  4. Pull in brand: “Hey Claude Code, where are my brand guidelines? Can you find it for me in my folders?” Claude Code searches the filesystem, locates SFCA brand guidelines in ~/Downloads/, reports the path. Now feed that file in.
  5. Second pass: branded PDF. Speaker says: “Within the first try it did this. Pretty solid.”

Two amplifiers Claude Code is using under the hood:

  • A “design skill” the speaker calls “a buff” — a downloaded skill defining graphic-designer rules. “When I’m telling it to design something, it knows: first I will check if we have a design skill that you buffed me with.” This is the standard Claude Code skills mechanism (see Skill Design Patterns and Agent Skills Overview). Speaker doesn’t show how to build the skill — points viewers to other YouTubers (Nick Saraev, Nate Herk).
  • Filesystem awareness — Claude Code can locate brand-guideline files on disk by name. Saves the “where did I save that?” hunt.

Notable un-prompted move: Claude Code named each of the 12 headline templates (“The Classic How-To,” “The Quick Win,” “The Mistake Callout”) even though the Google Doc just numbered them 1-12. Speaker comment: “It will often do more for you than you asked for, which is often good, sometimes screws things up.”

Phase 3 — Triple lead magnet (Claude Code + ChatGPT custom GPT)

Same content, three formats, one offer:

FormatToolWhat it is
Google DocoriginalEditable, shareable via link
Branded PDFClaude CodeStatic, downloadable, has paid-offer CTA at the end
Custom GPTChatGPT”AI headline coach” — paste your copy, get headline ideas using the 12 frameworks

Custom-GPT setup is a 5-minute job: ChatGPT → Create → upload the same DOCX, name it “Juicy Hooks Headline Coach,” set “Anyone with the link” sharing, copy URL, link from the PDF.

Why three formats: different audiences prefer different surfaces. The GPT also functions as a content lead magnet and a tool — speaker calls it a “triple whammy.”

The CTA at the end of the PDF (auto-generated by Claude Code from Brandon’s school.com about page): button + value-stack copy that pulls people from “free download” → “$10,700 in bonuses, join the academy.”

Phase 4 — Email automation (Beehiiv)

Beehiiv automation flow:

  • Trigger: “added by API” (from Zapier — see Phase 5)
  • Single email delivered: subject “Here’s your headline templates download link. Plus, I’ve got something for you.”
  • Email structure: short hook copy → branded graphic with link to the lead magnet → paid-offer pitch (~70% of the email body)

Speaker’s stance on email volume: one delivery email is fine for now. “You can do a full email series… reminder, hey don’t forget dive into your 12 Juicy Hooks.” But the demo ships one. He notes Beehiiv recently added A/B testing for automations, doesn’t use it in the demo.

The branded graphic for the email: Claude Code generates a “lead magnet on a fake monitor” visual by being shown an existing one and asked to recreate. First output is rough. “Make it more centered, make the margins wider, make the background transparent (not white).” Five tweaks later it’s good enough. Iteration honest-comment: “AI will get you 70-80% of the way there. Then you need to learn how to get it to 90-95%. Is it the absolute peak ability of top graphic designers? Not yet.”

Phase 5 — Landing + thank-you pages (Kajabi + Zapier)

Why two platforms: Brandon prefers Beehiiv for email but Kajabi for landing pages, so Kajabi → Zapier → Beehiiv. Trade-off honestly stated: “Things would be less complicated if I just kept it in Kajabi.”

Kajabi setup is rote — duplicate a previous landing page, change the URL slug, update the headline, swap the graphic, point the form to a new collector named “12 Juicy Hooks”:

  • Landing page copy structure: headline + sub-headline + form + button copy
  • Thank-you page: same pattern + “Click your link below” + paid-offer pitch (academy)
  • Form fields: email only (single opt-in, no name field — “I really don’t care”)

Zapier glue:

  • Trigger: Kajabi form submission on the “12 Juicy Hooks” form
  • Action: Beehiiv “Add subscriber to automation” with the matching automation ID
  • Test live by submitting your own email on the production form

Speaker bookmarks every URL into a browser folder named “12 Juicy Hooks Lead Magnet”: editable landing page, live landing page, editable thank-you page, live thank-you page, Zapier zap, Beehiiv automation. “I can quickly go to these links in case anything’s broken.” Suggests a Google Doc as an alternative when handing off to a client.

Phase 6 — Distribution (short video + Twitter + ManyChat)

The hand-off from “I have a funnel” to “I have leads” is one short-form video pushed across Instagram / Twitter / YouTube Shorts, plus a Twitter post with the graphic, plus a ManyChat Instagram automation.

Short-form video pipeline:

  1. Already filmed a 2-minute video pitching the headline guide (David Ogilvy quote → “comment HEADLINE for the free guide”).
  2. Captioning via Vizard.ai — upload the video, turn off “AI clips” (don’t want auto-trim), export with auto-captions. Captions matter because some people watch silent + better engagement.
  3. To generate the matching social-post text copy, find your own past winners on the same theme and feed them into Claude as templates. “I did a lead-magnet giveaway on X in the past and it did pretty well. Use this as a template for my post today.”
  4. Paste in (a) the past-winning post, (b) the new short-form video transcript, (c) the lead-magnet content. Ask Claude for matching post copy.

Distribution honesty: “Do I really need to use AI to rewrite this short of a post? Probably not.” He uses AI here to demo capability, not because it saves time on a 280-char post.

ManyChat Instagram automation — the biggest leverage move in Phase 6:

  • Trigger: anyone comments anything on a specific post
  • Reply 1 (public comment): “Thanks, check your DMs for the free juicy hooks copy templates”
  • DM 1: “Hey friend, thanks for commenting. Click the button below and I’ll send you my free 12 Juicy Hooks headline templates and AI headline generator.”
  • User taps the button (a yes/no interaction step that drives engagement signals)
  • DM 2: “Thanks, click the link below to grab your free templates” + landing-page link
  • DM 3 (follow-up if no click): another nudge

Three multipliers ManyChat unlocks:

  1. Email-list growth — automated capture without manual DMing
  2. Algorithmic boost — the comment trigger forces comments, which Instagram interprets as engagement, which expands organic reach
  3. DM conversation surface — opens a real-time sales channel

Speaker also demos manual DM delivery for Twitter (he used to use TweetHunter automation; “people were getting their accounts shut down so I now manually deliver in the DMs”).

The “AI Copywriter as Data Synthesizer” Framing

This is Brandon’s mental model for AI-assisted copywriting. Three sub-skills, in order:

  1. Extract data — know where the juiciest signal lives (community questions > non-buyer comments > competitor/Reddit). This is human judgment; AI can’t tell you where to look.
  2. Push data into the model — paste the right context plus the right prompt. “Hey Claude, I want you to give me 10 angles and ideas for lead magnets… here’s the data.”
  3. Pull data out of the model — refine via feedback loops, swap in your taste, chain-prompt to deepen (“now go build idea N”). This is also where format and design tweaks happen (“make the margins wider”).

Practical implication: spending 30 minutes scraping community questions in step 1 will out-perform any prompt-engineering effort in step 2. The data is the leverage.

Notable Patterns

  • CTA at every stage. Lead magnet PDF → email → landing page → thank-you page → DM → ManyChat sequence — every touch points to the paid offer. “Most people stop at: here’s the freebie, hopefully you pay me one day. No. Give them the next step right away.” Self-described as “why I’ve made my money as a copywriter.”
  • Repurpose, don’t reinvent. Same copy block reused across PDF intro, email body, thank-you page, social post copy. Same paid-offer pitch reused from prior lead magnet’s email automation. Same landing-page template duplicated. “Don’t work extra hard if you don’t need to.”
  • Reverse-engineer the funnel. Build last-step-first. End goal is customer → from a customer they came from → an email automation → which fired from a landing page → which got traffic from social. Each step exists for the next.
  • Test the funnel end-to-end with your own email before launching. “Always test your funnel.”
  • Bookmark folder per funnel. One browser folder, every URL inside, in funnel order. Edit-version + live-version of every page. “I can quickly go to these links in case anything’s broken.”

Tool Stack (as demoed)

ToolRole
Claude (web)Ideation: turn audience signal into 10 lead-magnet angles
Claude CodePDF production, branded graphic, find-files-on-disk, generate CTA copy
ChatGPT (custom GPT)Triple-lead-magnet third format (“AI headline coach”)
WhisperFlowVoice-to-text for prompting Claude
FathomCoaching-call recording + transcript extraction
BeehiivEmail service + automation
KajabiLanding page + thank-you page (alt: Go HighLevel as all-in-one)
ZapierKajabi form → Beehiiv API bridge
Vizard.aiAdd captions to short-form video
ManyChatInstagram comment → DM automation
school.comCourse/community hosting (not Claude-related — the destination of the funnel)

Try It

If you’re a copywriter or marketer at WEO Marketly with a paid offer or course in hand, the leanest version of Brandon’s playbook to apply this week:

  1. Pick the lead magnet from existing material. Don’t ideate from scratch — break a piece off something you already have (course module, internal SOP, a one-pager you sent a client). Brandon’s worked example is literally the “12 headline templates” page from his paid academy.
  2. Run Phase 1 in 20 minutes. Open Claude (web). Paste 50-100 audience questions / comments / coaching-call transcript snippets. Ask for 10 lead-magnet angles. Pick one with your taste, not Claude’s.
  3. Use Claude Code only for the production step. Drop the source doc as PDF → “make me a pretty PDF version branded for X audience.” Iterate conversationally. Add a CTA at the end by feeding it your sales page or about page. Stop at “good enough” — Brandon explicitly gives up at ~85% on visual polish.
  4. Optional but recommended — make a custom GPT version of the same content. Five minutes in ChatGPT, instant upgrade from “PDF” to “AI tool” framing.
  5. Bookmark every URL into one folder before launch. Editable + live versions of: landing page, thank-you page, automation, Zap, social posts. This is the unglamorous habit that saves the hour-long debug when something breaks at 11pm.
  6. Test with your own email end-to-end. Submit through the production form. Confirm Zap fires, automation triggers, email arrives, link works, thank-you page redirects.
  7. For the WEO context: the marketing automation use cases catalog already lists adjacent patterns; this Brandon-style playbook fits as the “lead magnet” entry point of any of them. Apply the 6-question vetting framework before downloading the “design skill” Brandon references.

Implementation

Tool/Service: Claude Code with a downloaded design skill, plus the multi-tool stack above.

Setup notes:

  • Claude Code can’t ingest DOCX directly — convert source docs to PDF first (PDF / PNG / JPEG / GIF supported).
  • The “design skill” is referenced but not built in this video. The standard skill format applies (see Agent Skills Overview); design-specific examples are in VoltAgent’s awesome-design-md and the wider design-skills-overview.
  • The “find brand guidelines on my computer” move requires Claude Code with file-system access — Computer Use isn’t needed for this; standard Read/Bash tooling handles it.
  • For graphic generation, Claude Code’s first output is reliably ~70-80% of target. Plan for 4-6 iteration turns to get to “good enough.”

Cost: Free side — Claude (web), ChatGPT custom GPT, school.com community paid only. Paid stack: Beehiiv + Kajabi + Zapier + Vizard.ai + ManyChat. Speaker doesn’t disclose totals; rough order of magnitude $200-400/mo combined for a working solo-creator funnel.

Integration notes:

  • Auto-caption mishearings flagged for future search: “Beehive” → Beehiiv; “Glass” → likely Glasp (the YouTube-summarizer Chrome extension); “Nick Sarive” → likely Nick Saraev; “Nate Hurk” → Nate Herk; “Cloud Anti-Gravity” appears once in transcript and is unclear — possibly an accidental name-mash; the speaker just opens Claude Code immediately after.
  • The video is a 1:55:54 free course functioning as a lead magnet for Brandon’s paid Six Figure Copy Academy. Standard lead-magnet-funnel discipline applies — the value is real, but the recommendations skew toward solo-creator scope and DTC/freelancer use cases. WEO-Marketly multi-client agency context will need adjustments (most obviously: per-client brand guidelines vs. a single house brand).

Open Questions

  • The “design skill” is described as “downloaded from online” but never named or linked. What’s the actual source? Speaker points to Nick Saraev / Nate Herk for skill-building tutorials — those creators have public skills repos worth surveying.
  • The triple-lead-magnet pattern (Doc + PDF + GPT) is presented as obviously valuable, but no conversion data is shown comparing single-format vs. triple-format pickup rates. Worth A/B testing in a WEO context before assuming the GPT step earns its setup time.
  • Brandon’s funnel pushes the paid offer at every stage (PDF + email + landing + thank-you + DM). For a regulated vertical like dental marketing, would this density of “buy now” CTAs trigger compliance issues (state advertising boards, FTC endorsement rules)? Worth a pass against WEO AI policies before adapting wholesale.
  • Beehiiv’s API trigger via Zapier (Phase 5) is a reliable but fragile pattern. What’s the failure mode rate? Speaker doesn’t show monitoring or alerting on broken Zaps.
  • The “data synthesizer” framing maps neatly to RAG architectures — is there a meaningful upgrade path from Brandon’s manual paste-into-Claude pattern to a per-client brand-context skill that lives in a connected-skills-as-agentic-OS style architecture?

Graph View

Backlinks