Jonathon's AI Wiki

Module 3 — Role Playbooks by Discipline

26 min read

Source: wiki synthesis: marketing-skills-bundle, AI Marketing topic; course-original prompt examples. RCTF pattern from Module 2.

Time: Read 15 min (your role section only) | Practice 30 min

The WEO value-add of this course. Five role-based playbooks with 25 before/after prompt pairs — the naive version almost everyone writes first, the RCTF version that actually works, and a one-line explanation of what changed. Jump to your role. Come back for the others as you need them.

How to use this module:

  • Each prompt has three parts: Scenario, Before (what most people write first), After (the RCTF version), and Why it wins (what changed).
  • Copy the After prompt. Fill in placeholders like [PRACTICE NAME] with real values.
  • Examples are dental because WEO is a dental agency. Patterns transfer to any vertical.
  • Governance reminder: don’t paste real patient data, real client PII, or unreleased strategy into Claude while practicing. Module 7 covers what’s allowed on real work.

Claude 101 cross-reference

Anthropic’s Claude 101 course has its own “Claude in action: use-cases by role” section — worth skimming after you finish your WEO role section here. It covers roles beyond marketing (product, engineering, operations) and shows how the same prompting patterns generalize.

Community walkthroughs (optional extras)

For video learners — two business-user-focused walkthroughs aimed at online-business / marketing-agency staff (not developers). Both from Rick Mulready, already cited in Module 8:

For WEO’s specific content-generation workflow, Nate Herk’s “Generate Content for 9 Socials on Autopilot with Claude Code” shows what engineering-adjacent roles can automate.

Hands-on Cowork recipes (verbatim prompts you can lift)

When you finish your role section and want to feel Cowork running on real work, three external Notion AI Recipes ship with copy-paste prompts that map straight to roles in this module. All three live in a curated “AI Recipe Vault” Notion catalog:

  • Getting Started with Claude Cowork — 7-step quickstart + 5 non-obvious use cases each shipped as a complete prompt: The Hiring Manager (raw CVs → leadership-summary spreadsheet — pairs with the Customer Support / role-flow material in 3.1), The SEO Auditor (GSC export + Chrome browse → audit report — direct fit for Section 3.4 SEO), The Customer Intelligence Builder (Gmail folder → ICP report + HTML dashboard — fits 3.5 Content / 3.4 SEO discovery), The Content Strategist (Notion competitor list → 30-day content plan back into Notion — pairs with 3.5), and The Group Travel Agent (chaotic bookings → shareable itinerary — general-purpose). Use these as your second Cowork session after the file-org starter.
  • Cowork + Apify Scraping Recipe — 6-step prospect-list workflow with verbatim prompts at each step. Fits the SEO / Ads / sales-adjacent paths in 3.4 — local business scraping → vibe prospecting → LinkedIn job-description scrape → PR-ready AI-skills research report. Apify free tier covers the recipe end-to-end.
  • The LinkedIn Engagement Machine — 4-prompt LinkedIn content chain. Maps to 3.5 Content. Same-thread context-retention is the architectural lesson — runs all four prompts in one chat so the model retains the input transcript.

All three caveats: source pages are UK-authored Notion templates (British spelling), not Anthropic-published docs. Treat as starting frameworks, not best-practice claims; pair with Module 7’s governance rules before running on real client data.

3.1 Customer Support

Claude is excellent at drafting first-pass responses, softening tone, summarizing long email threads, and turning scattered information into clean status updates. Support pros stay in the loop — Claude drafts, you edit and send.

Prompt 1 — Draft a response to a client complaint

Scenario: Client is frustrated because a blog post was delivered two days late due to a scheduling-tool failure. Tool is fixed, blog is live, you want to offer a 10% credit as goodwill.

Before (what most people write first)

Write an email to a client apologizing for a late blog post and offering a 10% credit.

After (RCTF)

Role: You are a customer success manager at a dental marketing agency with a
calm, empathetic, solution-oriented communication style.

Context: A client is frustrated because their blog post was delivered two days
late. The delay was caused by our scheduling tool failing. We've already fixed
the tool, the blog is now live, and we want to offer a 10% credit on next
month's invoice as a goodwill gesture.

Task: Draft an email to the client acknowledging the issue, explaining briefly
what happened (without blaming the tool — own it), confirming resolution, and
offering the credit.

Format: 150 words max. Warm but professional. Subject line + body. No
"we apologize for any inconvenience this may have caused" — too corporate.

Why the after wins: The Role sets empathy + ownership. The Context gives Claude the real story (tool failure, resolution, credit amount) so it writes accurately. The Format constraints — word count, structure, banned phrases — prevent corporate-boilerplate output.

Prompt 2 — Summarize a long email thread

Scenario: 30-message client email thread you need to get up to speed on quickly.

Before

Summarize this email thread. [paste]

After (RCTF)

You are my executive assistant. I'm about to paste a long email thread
between me and a client. Read it carefully and give me:

1. A 3-bullet summary of what's happened so far
2. The outstanding question(s) the client is waiting on me to answer
3. One suggested next action for me

Keep it under 150 words total. Here's the thread:

[PASTE THREAD]

Why the after wins: The “executive assistant” role gets Claude into a “serve me, don’t over-explain” posture. The three-part structure forces a decision-ready output (not just a summary). The next-action item is what actually unblocks you.

Prompt 3 — Soften an angry draft before sending

Scenario: You wrote an email in frustration and caught yourself before sending.

Before

Make this email nicer. [paste]

After (RCTF)

I wrote this draft reply to a client when I was frustrated. I don't want to
send it as-is. Rewrite it to preserve my actual points but sound like a
calm, reasonable professional. Don't capitulate — the client is wrong on
the facts — but strip out the emotional tone.

Draft:
[PASTE YOUR ANGRY DRAFT]

Why the after wins: Explicitly names “don’t capitulate — the client is wrong on the facts.” Without this, Claude defaults to overly-apologetic rewrites that concede your position. You want composure, not surrender.

Prompt 4 — Generate an FAQ answer

Scenario: Adding an FAQ about SEO timelines to the agency’s website.

Before

Write an FAQ answer about how long SEO takes to work.

After (RCTF)

Role: A knowledgeable customer support rep who writes helpful FAQ answers.

Context: We're a dental marketing agency. One of our FAQ page questions is:
"How long does it take to start seeing results from SEO work?"

Task: Write a 2-paragraph answer that sets realistic expectations (3-6 months
for meaningful movement), acknowledges the variability by market/competition,
and ends with an action-oriented sentence inviting them to talk to us about
their specific timeline.

Format: Conversational, no jargon. Avoid guaranteeing specific outcomes — SEO
results depend on many factors.

Why the after wins: Specifies the 3-6 month honest window (Context), bans outcome guarantees (compliance-aware), and directs toward a CTA instead of a dead-end explanation (Task). This is the difference between generic FAQ content and WEO’s actual voice.

Prompt 5 — Escalation email to a manager

Scenario: Client has raised the same concern 3 times; you need your manager to intervene.

Before

Write an email to my manager explaining that a client is frustrated.

After (RCTF)

Role: A support rep drafting an internal escalation email to my manager.

Context: A client has now raised the same concern (unclear monthly reporting)
three times over two months. I've escalated twice informally and it hasn't
stuck. I need my manager to intervene.

Task: Draft an internal email to my manager [NAME] laying out:
- What the client has asked for (3 times)
- What we've done in response (or failed to do)
- Why I think it needs her attention now
- What I'd propose as next steps

Format: Clear, not emotional, not blame-y. Short paragraphs. 200 words max.

Why the after wins: The four-part Task forces structure (what / history / why now / proposal) — a manager can act on this, not just read it. “Not blame-y” in Format prevents the tone from sliding into finger-pointing, which escalation emails do by default.

3.2 Web Design

Claude is a strong copy and content partner for web designers — drafting hero sections, service page narratives, microcopy, and content briefs. It won’t design the layout, but it’ll fill the rectangles with words that don’t suck.

Prompt 1 — Homepage hero copy

Scenario: New client site; you need 3 hero variations for design review.

Before

Write me a headline and subhead for a dental practice website.

After (RCTF)

Role: Senior website copywriter specializing in dental practice marketing.

Context: Practice name: [NAME]. Location: [CITY, STATE]. Target patients:
[age range, demographic]. Key differentiator: [e.g., "pediatric specialty,"
"family-friendly with sedation options," "top implant specialist in county"].
Brand voice: [e.g., warm and plainspoken, premium and understated, modern
and playful].

Task: Write 3 variations of a homepage hero section. Each should include:
- A headline (max 8 words)
- A subheadline (max 15 words)
- A single CTA button label (2-4 words)

Format: Clearly numbered (1, 2, 3) with each element labeled.

Why the after wins: Naming differentiator and brand voice is the difference between generic dental copy and “this sounds like THIS practice.” Word-count constraints make the output actually fit the design — headlines that read great at 18 words don’t fit a hero at 320px.

Prompt 2 — Service page structure + copy

Scenario: Building out an Invisalign service page for a client.

Before

Write the content for an Invisalign service page for a dental website.

After (RCTF)

Role: Website content strategist for healthcare marketing.

Context: I'm building a service page for [PRACTICE NAME]'s Invisalign offering.
Target patients: adults 25-45 looking for discreet orthodontics. Competitors'
pages are mostly clinical — we want ours to feel more consumer-friendly.

Task: Propose a complete service page structure with H2 sections and 2-3
sentence content for each. Sections should cover: overview/benefits,
how it works (process), who's a good candidate, how it compares to braces,
cost/financing angle (general, no specific numbers), and a clear CTA to book
a consultation.

Format: H2 headings + body copy under each. About 400-500 words total.
End with 3 suggested CTAs for button variants.

Why the after wins: Naming the competitor posture (“mostly clinical — we want consumer-friendly”) steers voice. Listing the six required sections prevents omissions. Asking for 3 CTA variants gives design options without a second prompt.

Prompt 3 — Microcopy for a form

Scenario: Consultation request form for a family dental practice.

Before

Write the text for a consultation request form on a dental website.

After (RCTF)

I'm designing a "Request a Consultation" form for a dental practice website.
Write me warm, non-clinical microcopy for every element:

- Form heading
- Form intro sentence (1 line, sets expectations for what happens after submit)
- Field labels and placeholders for: Name, Phone, Email, Preferred appointment
  time, What brings you in today (optional)
- Submit button label
- Success state message
- Error state message (generic validation failure)

Keep it human. No "please fill out all required fields." The practice
is family-friendly, midwest, middle-income target.

Why the after wins: Naming every single microcopy element (8 of them) makes sure nothing gets skipped. Banning “please fill out all required fields” specifically kills the worst default phrase. Adding demographic context (midwest, family-friendly) steers register.

Prompt 4 — Content brief for a designer

Scenario: Redesigning an “About Us / Meet the Doctor” page that’s currently dry.

Before

Write a brief for an About Us page redesign.

After (RCTF)

Role: Marketing strategist producing a content brief for our design team.

Context: We're redesigning the "About Us / Meet the Doctor" page for
[PRACTICE]. Current page is dry, resume-style. We want the new version to
feel personal and build trust. The doctor is [BRIEF BIO].

Task: Write a content brief for the designer covering:
- Goal of the page (in one sentence)
- Target emotional response when a prospective patient finishes reading
- Must-have sections (in order)
- Tone guidance
- 3 specific do's and 3 specific don'ts
- Word count target

Format: A clean, scannable brief a designer can work from.
About 300 words. Use headers and bullet points.

Why the after wins: “Target emotional response” is a design-brief technique that lifts work quality. Explicit do’s/don’ts prevent the designer guessing. 300-word cap prevents a wall of text a designer won’t read.

Prompt 5 — Rewrite designer copy to sound human

Scenario: Existing homepage reads like every other dental site.

Before

Make this sound less generic. [paste]

After (RCTF)

Our homepage currently reads like every other dental practice homepage.
Rewrite this section to sound like a real human at a family-friendly
practice talked to us, not like template text.

Keep the same information. Change the feel. Remove clichés like
"state-of-the-art" and "world-class." Target 25% fewer words.

Current copy:
[PASTE COPY]

Why the after wins: Naming the specific clichés to kill is more effective than abstract “make it less generic.” The 25% word-count reduction forces tightening — Claude will pad if you don’t.

3.3 SEO

Claude accelerates the research-and-write parts of SEO: keyword clusters, outlines, metadata, FAQ schema ideas, content briefs. It does NOT replace a real SEO tool (SEMrush, Ahrefs, GSC) — always ground Claude’s keyword ideas in real search data before you ship.

Prompt 1 — Keyword cluster brainstorm

Scenario: Building out a “Dental Implants” service page + supporting blog cluster.

Before

Give me some keywords for a dental implants page.

After (RCTF)

Role: SEO strategist specializing in local dental marketing.

Context: Client is a general dentist in [CITY, STATE]. We're building out
content for their "Dental Implants" service page and a blog cluster around it.

Task: Propose a keyword cluster with:
- 1 primary target keyword (the service page itself)
- 5 secondary keywords (supporting blog topics)
- 3-5 long-tail "question" keywords per secondary topic

Focus on commercial and informational intent. Do NOT invent search volumes —
I'll validate everything in Ahrefs. Mark each keyword with my best guess at
intent (informational, commercial, transactional).

Format: Nested bullet list.

Why the after wins: Explicitly banning invented search volumes stops Claude from hallucinating traffic numbers — a real problem in SEO work. The nested primary/secondary/long-tail structure is a real keyword strategy shape, not a flat keyword list.

Prompt 2 — Meta description

Scenario: Writing meta descriptions for a batch of service pages.

Before

Write a meta description for a teeth whitening page.

After (RCTF)

Write 3 variations of a meta description for this page. The page is
[URL or topic]. Target keyword: [KEYWORD]. The page's main benefit for a
searcher is [BENEFIT].

Constraints:
- Exactly 150-160 characters (I will count)
- Include the target keyword naturally
- End with a clear CTA or benefit hook
- Avoid starting with "Welcome to" or "Looking for"

Output as a numbered list with character counts after each.

Why the after wins: “I will count” is a tiny phrase that measurably improves Claude’s output on character constraints. Banning the two worst openers kills 80% of weak meta descriptions in one line. Variants let you A/B in GSC.

Prompt 3 — FAQ schema ideas

Scenario: Adding FAQ schema to a new dental-implants service page.

Before

Write some FAQs about dental implants.

After (RCTF)

Role: SEO content strategist focused on structured data.

Context: I'm adding FAQ schema to a service page about [SERVICE, e.g., dental
implants]. Target audience: patients researching the procedure for the first
time.

Task: Generate 8 FAQ questions + 2-3 sentence answers. Prioritize questions
real patients search for (cost, pain, recovery, candidacy, insurance).
Answers should be accurate, compliant (no outcome guarantees), and
schema-ready.

Format: Q1/A1, Q2/A2, etc. Each answer should be 30-60 words.

Why the after wins: Explicitly naming the 5 real-world patient concerns (cost/pain/recovery/candidacy/insurance) ensures the FAQs match actual search intent. “Compliant, no outcome guarantees” is the dental-marketing safety rail Claude would otherwise ignore.

Prompt 4 — Content outline from a keyword

Scenario: Planning a “teeth whitening vs. veneers” comparison post.

Before

Write an outline for a blog post comparing teeth whitening and veneers.

After (RCTF)

Role: SEO content strategist.

Context: I want to write a blog post targeting the keyword "teeth whitening
vs veneers." The searcher intent is comparison — someone trying to decide.
The client is a cosmetic dentist in [CITY].

Task: Propose a complete blog outline:
- Suggested title (1 main + 2 alternates)
- Meta description (draft)
- H2 and H3 outline with 1-sentence summary of each section
- Suggested word count
- 3 internal linking opportunities (to typical dental practice pages)
- 3 schema types that would benefit this page
- Rough word count estimate

Format: Structured, scannable.

Why the after wins: Naming searcher intent (“comparison — someone trying to decide”) steers the whole outline toward decision-support, not general explanation. Internal links + schema recommendations turn this from “an outline” into “a complete content plan.”

Prompt 5 — Competitor SEO teardown summary

Scenario: Analyzing a competitor’s top-ranking page before you out-rank it.

Before

Analyze this article and tell me what's good about it. [paste]

After (RCTF)

I'm about to paste the top-ranking URL for "[keyword]" in [city]. Read the
content and give me:

1. Main topics covered (bulleted)
2. Content gaps — what they did NOT cover that a reader would want
3. 3 angles I could use to create a stronger page
4. Rough word count estimate
5. Tone/voice description

[PASTE COMPETITOR CONTENT]

Why the after wins: The naive prompt gets flattery. The RCTF version explicitly asks for gaps (what they missed) — that’s where your competitive edge comes from. The 5-part structure is a teardown template you’ll reuse weekly.

3.4 Paid Ads

Claude is a variant machine — give it one good seed ad and it’ll generate 20 viable variations. It’s also a solid sparring partner for audience angles, hook ideas, and A/B hypotheses. Always pair Claude-generated ad copy with platform-specific character limits and compliance checks.

Prompt 1 — Google Ads RSA headlines

Scenario: Launching a new Invisalign campaign for a practice.

Before

Write me 15 Google Ads headlines for Invisalign.

After (RCTF)

Role: Paid search copywriter for dental practices.

Context: Client: [PRACTICE], [CITY]. Service: [SERVICE, e.g., Invisalign].
USP: [key differentiator]. Landing page: [URL or 1-line description].
Location target: 10-mile radius around practice.

Task: Write 15 distinct headlines for a Responsive Search Ad (RSA). Each
MUST be 30 characters or fewer (I will count). Mix angles:
- Benefit-led (3)
- Location-led (3)
- Question/curiosity (3)
- Price/offer-led (3, no specific numbers — use "affordable," "transparent")
- Trust-led (3, e.g., "Board-certified," "20+ years")

Format: Numbered list with character count after each. Reject any that
don't fit.

Why the after wins: The naive prompt returns 15 similar headlines. The RCTF version returns 15 headlines across 5 distinct angle-types — which is what Google’s RSA machine-learning needs to actually A/B. Character count discipline prevents Google Ads truncation.

Prompt 2 — Meta primary text variants

Scenario: Diversifying ad creative on a campaign with flat CTR.

Before

Write me some Facebook ad copy for the same Invisalign campaign.

After (RCTF)

Role: Direct-response Meta ads copywriter.

Context: [SAME CONTEXT AS ABOVE]. Audience: adults 25-45 on Facebook and
Instagram in [metro area]. Offer: free consultation.

Task: Write 5 primary-text variations for a Meta ad. Each should be
90-125 words, lead with a hook, build desire, and close with a specific
CTA to book. Variations should test different angles:
1. Problem-awareness (pain point)
2. Aspiration / transformation
3. Social proof angle
4. Urgency / limited availability
5. Curiosity / counterintuitive

Format: Labeled by angle (1–5). Include a one-line hypothesis after each
explaining why this variant might win.

Why the after wins: Asking Claude to label each variant by angle + attach a hypothesis forces it to produce different copy, not 5 minor rephrasings. The hypothesis line is what your ad manager uses to decide which to test first.

Prompt 3 — Audience angle brainstorm

Scenario: Campaign has flat CTR for 3 months — you need fresh angles.

Before

Give me some new ideas for my emergency dentistry ads.

After (RCTF)

Role: Paid media strategist for dental services.

Context: We're advertising [SERVICE, e.g., emergency dentistry] in [CITY].
Current campaign has been running for 3 months with flat CTR. I need fresh
angles.

Task: Propose 8 distinct audience angles we haven't tried. Each should include:
- Who the audience is (psychographic, not just demographic)
- The core pain point / desire
- The creative angle (hook or message)
- Why this angle is different from what we've been running

Format: Numbered list, 2-3 sentences per angle.

Why the after wins: “Psychographic, not just demographic” forces Claude past the standard “adults 35-55” output into actual segments (parents with a midnight-crying kid, a young professional who can’t take time off, etc.). The “why this is different” field prevents repetition.

Prompt 4 — A/B test hypothesis

Scenario: You want to propose 3 tests to your account manager with rigor.

Before

Suggest 3 A/B tests for my ad campaign.

After (RCTF)

Role: Growth strategist with a strong experimental mindset.

Context: Current ad's CTR is [%], CVR is [%], CPA is [$X]. Ad copy:
[CURRENT COPY]. Landing page copy: [CURRENT LP HEADLINE]. Audience: [DESC].

Task: Propose 3 A/B test hypotheses. Each should:
- State the hypothesis in "If we change X, Y will happen, because Z" format
- Specify the exact variable(s) to change
- Predict the primary metric that will move
- Suggest a success threshold (e.g., "+15% CTR")
- Rank the three by expected impact × effort

Format: Structured per hypothesis. Be rigorous — don't suggest vibes-based
changes.

Why the after wins: The “If X, Y, because Z” format is real experimental-design language — it forces a testable mechanism, not “let’s try this.” Ranking by impact × effort gives the account manager a decision, not a list.

Prompt 5 — Negative keyword suggestions

Scenario: Reviewing a week of search terms for a client’s campaign.

Before

Look at these search terms and tell me which to add as negative keywords. [paste]

After (RCTF)

I'm running Google Ads for [SERVICE] in [CITY]. I'm about to paste my current
search terms report (the queries that triggered my ads). Identify:

1. Queries that are clearly irrelevant (should be negated)
2. Queries that are ambiguous (worth watching but not yet negating)
3. Queries that reveal a new intent worth building a new ad group around

For each negative recommendation, specify exact match vs. phrase match.

Search terms report:
[PASTE]

Why the after wins: Adding “watch vs. negate” as a middle category prevents premature negation (real strategic mistake). The “new ad group” angle turns search-term reports from defensive into offensive — finding campaign expansion opportunities.

3.5 Content Marketing

Claude is a force multiplier for content — outlines, drafts, repurposing, editing, headline variants. The unlock is using Claude for the 70% that’s mechanical so you spend your time on the 30% that makes content actually good (judgment, perspective, voice).

Prompt 1 — Blog outline from a topic

Scenario: Planning a new blog post for a pediatric dental client.

Before

Write an outline for a blog post about kids' dental anxiety.

After (RCTF)

Role: Senior content strategist for dental marketing.

Context: I want to write a blog post for [CLIENT, practice type]. Topic:
[TOPIC]. Target reader: [AUDIENCE]. Goal: [e.g., rank for "X keyword," build
trust with prospective patients, support a specific service page].

Task: Propose a complete outline:
- 3 title variations
- Suggested meta description
- Intro hook (1 paragraph, actual copy — not just a summary)
- H2 sections with 2-3 sentence summaries of each
- Suggested word count
- 2-3 internal links (to typical practice pages like service pages, about page)
- 1 clear CTA for end of post

Format: Structured, scannable, ready to hand to a writer.

Why the after wins: “Goal” field forces the post to serve a measurable outcome (rank, trust, support) instead of being generic education. Asking for actual hook copy, not a summary, means you can see if the post has a voice before you commission it.

Prompt 2 — Full blog draft from an approved outline

Scenario: Turning a approved outline into a first-draft blog post.

Before

Write a blog post based on this outline. [paste]

After (RCTF)

Role: Dental marketing copywriter. Voice: warm, plainspoken, trustworthy.
Not clinical.

Context: [CLIENT, practice type, city]. Target reader: [AUDIENCE]. Blog
topic: [TOPIC]. The outline below is approved — follow it closely.

Task: Write the full blog post. Match word counts per section. Use specific,
tangible examples instead of abstractions. Include one patient-story
illustration (fictional but realistic). End with the CTA in the outline.

Format: Full draft in markdown (H2/H3 headings, paragraphs, occasional
bullet points where they help). Target word count: [NUMBER]. Flag any claim
that should be fact-checked with a clinician before publishing by wrapping
it in [VERIFY: ...].

Outline:
[PASTE OUTLINE]

Why the after wins: “Voice: warm, plainspoken, not clinical” + “tangible examples instead of abstractions” are the two biggest levers for dental-marketing content quality. The [VERIFY: ...] flag is a WEO-specific safety pattern — it keeps clinical hallucinations from reaching publication.

Prompt 3 — Repurpose a blog post into social

Scenario: Just published a blog — now need 2 weeks of social content from it.

Before

Turn this blog post into social media posts. [paste]

After (RCTF)

I just published the blog post below. Repurpose it into:
- 3 LinkedIn posts (each standalone, 150-200 words, conversational)
- 5 Instagram captions (each 100 words max, emoji-friendly, one clear hook
  in the first line)
- 10 tweet/X posts (each under 280 characters, varying angles —
  question, stat, surprising take, advice, quote)
- 1 email newsletter teaser (3 paragraphs, links back to the post)

Each should stand alone — assume the reader doesn't click through to the blog.

Blog post:
[PASTE BLOG]

Why the after wins: Platform-specific formatting + word counts mean the output is paste-ready, not “needs a second editing pass to fit the channel.” “Each should stand alone” prevents the common mistake of writing teasers that only make sense if you’ve read the blog.

Prompt 4 — Newsletter subject line variants

Scenario: Monthly patient newsletter, you want 12 subject lines to test.

Before

Write some subject lines for a dental newsletter.

After (RCTF)

Role: Email marketer with strong subject-line instincts.

Context: Monthly patient newsletter for [PRACTICE]. This month's main content
is [TOPIC / THEME]. Audience: existing patients, all ages.

Task: Write 12 subject line variations across these angles:
- Curiosity (3)
- Benefit-led (3)
- Personal/warm (3)
- Practical/tip-led (3)

Constraints: 50 characters max each. No clickbait. No emojis. Avoid spam
triggers (FREE, ALL CAPS, etc.).

Format: Numbered list, grouped by angle, character count after each.

Why the after wins: Grouping by angle gives you 4 different A/B candidates, not 12 variants of the same angle. The spam-trigger exclusion is a practical deliverability win most people forget to ask for.

Prompt 5 — Topic ideation session

Scenario: Planning a quarter’s worth of content for a client.

Before

Give me 12 blog topic ideas for a dental practice.

After (RCTF)

Role: Content strategist running a content planning session with me.

Context: I need to plan the next 3 months of blog content for [CLIENT].
The practice offers [LIST OF SERVICES]. Their main SEO gaps are around
[TOPICS]. They want to build more local authority in [CITY].

Task: Propose 12 blog post ideas for the quarter. For each:
- Working title
- Target keyword / topic
- Primary reader (which patient segment)
- Which of the practice's services it supports
- Estimated impact (SEO traffic / trust-building / referral-worthy) —
  rate on 1-5
- Suggested publishing order and reasoning

Format: Table or structured list.

Why the after wins: Without “SEO gaps” and “services offered” in Context, Claude proposes generic dental-blog ideas (12 practices could publish the same list). With them, the output maps to this client’s strategic position. The impact rating gives you an argument for sequence — which post goes first.

Key Takeaways

  • RCTF structure every time. Every After prompt above follows it.
  • “Before → After → Why” is the pattern for teaching this. Use it when you teach a junior team member how to prompt.
  • Save the winners. When an After prompt produces work you’d ship, save the prompt template (with placeholders re-inserted) to your personal prompt library.
  • Claude drafts, you edit. Treat outputs as first drafts you improve, not final work.
  • Always verify clinical / regulatory / local claims before publishing on client work. Module 7 covers what’s allowed.
  • Cross-role learning. Scan other roles — SEO patterns help content pros, support patterns help account managers.

Try It (30 min, hands-on)

Pick your primary role and do these. If you wear multiple hats, pick the one where you’d get the most immediate value.

  1. Copy one After prompt from your role section. Fill in the placeholders with something real from your current work (sandbox data only — no client PII).
  2. Send it in claude.ai.
  3. Now send the Before version. Same chat, different thread. Compare the two outputs side-by-side. The difference between them is this course’s core lesson.
  4. Iterate using the 3 moves from Module 2 (steer, pivot, edit-and-resend) until the After output is something you’d actually use.
  5. Save the prompt to your notes or a Claude Project (see Module 4) — with placeholders re-inserted as {{TEMPLATE_VARIABLES}}. First entry in your personal prompt library.
  6. Now skim one role section that’s NOT yours. You’ll find at least one prompt you can adapt — support people benefit from content patterns, designers benefit from SEO patterns.

Done? Move on to Module 4.

Graph View

Backlinks