Source: X Thread Linkedin Funnel 7 Prompts 2026 04 19 — X/Twitter thread, author and source URL not provided by user. Attribution flagged as unknown. Lead-magnet-style thread (closes with DM-trigger CTA “Drop ‘SYSTEM’ in the comments…”).

A seven-prompt Claude workflow for an end-to-end LinkedIn lead-gen funnel: profile optimization → ICP pain-point research → weekly content system → sales-call mining → content repurposing → cold outbound → lead-magnet post. Every prompt (except the lead-magnet one) opens with “interview me in batches” and produces one concrete deliverable — a running theme across the other LinkedIn-prompt collections in this topic (Amy Edmondson thread, Seth Godin thread). Broader in funnel coverage than either — this one stretches from profile hygiene to cold DMs.

Key Takeaways

  • End-to-end funnel coverage, not just content. Profile (1) → research (2) → content calendar (3) → sales-call content mining (4) → repurposing (5) → outbound (6) → lead-magnet post (7). The Amy Edmondson 5-prompt thread covers narrative → signature idea → calendar → proof → comments strategy — overlaps on content, extends past it into outbound DMs and sales-call mining.
  • Interview-pattern as default. Six of seven prompts start with “Before writing anything, interview me in batches.” Batches are numbered. This matches the interview-pattern described in Prompt Engineering Essentials and used in all God of Prompt threads.
  • Prompt 7 breaks the interview pattern intentionally. The lead-magnet post prompt takes context inline (single Context: [...] block) and outputs the post directly in a code box with no explanation. Designed for reuse — you’ve already done the discovery in prompts 1–6; this is the distilled output prompt.
  • Ban-lists are unusually concrete. Prompt 7 bans specific phrases: em dashes, “not X not Y just Z” pattern, “let’s dive in,” “here’s the thing,” “game-changer,” “unlock,” “transform your X,” fragment hooks without context, generic “most [audience]” openings. This is Anti-AI Slop Guide applied to LinkedIn copywriting. Worth borrowing for any LinkedIn prompt.
  • Prescribed output structures. Each prompt specifies exact deliverable shape: WAR framework for headlines (What + Authority + Results), 7-post weekly calendar split 2/3/2 (top/mid/bottom of funnel), 30-day campaign spec snapshot, content map as minimum 14-row table, hook-reframe-credibility-tease-CTA post structure. Removes ambiguity.
  • Uses prospects’ own language as an explicit input. Prompts 2 and 4 both specify “in users’ own words” and “prospects’ actual language” — mining Reddit or sales-call transcripts. Reduces the “AI-sounding” quality that afflicts default LinkedIn output.
  • WAR headline framework is net-new vocabulary. Not in the existing LinkedIn articles in this topic. W = what I do, A = authority, R = results. Concrete mnemonic worth propagating to the Edmondson and Godin articles if they refresh. ^[inferred — “WAR framework” is asserted in the source without external reference; no corroborating source yet]
  • Syntax: conversational instructions, not structured tags. Unlike the Winston and Godin threads (XML tags) or the Edmondson thread (#ROLE: markdown headers), this thread uses plain prose with “Batch 1/2/3” and “Phase 1/2” as structural cues. Less parseable but more natural when a non-technical founder pastes into claude.ai.

The Seven Prompts at a Glance

Full prompt text in the source file. Summary:

#PromptGoalDeliverable shape
1Optimize your profilePosition as trusted authority in under-minute profile scan3 WAR headlines, featured section (website / best post / lead magnet), link CTA, experience rewrites, 2 skills to endorse, profile-pic checklist, banner direction, content cheat sheet — ready-to-paste
2Research ICP pain pointsSurface exact prospect language for hooksPain-point taxonomy (primary / secondary / emerging / unmet needs) with direct quotes + 1–10 intensity + frequency + current solutions + gaps + 5 LinkedIn posts for the week
3Weekly content systemStop guessing what to post7-post weekly calendar 2/3/2 (top/mid/bottom), all 7 posts ready-to-paste, bold hook, no hashtags, bullet points + line breaks, numerals not words
4Mine sales callsTurn transcripts into contentTop 7 objections + top 7 pain points + top 7 solutions (each in prospects’ words with frequency), 14-row content map, 7 LinkedIn drafts using prospects’ actual language
5Repurpose long-formOne asset → full week5–7 standalone micro-posts (mix of single-insight / contrarian / mini-story / listicle / proof / question / carousel outline), posting order with strategic reason, visual suggestion per post
6Outbound systemTargeting → booked calls30-day campaign spec, Sales Navigator filter instructions, ICP avatar doc, 3 cold DM variations (each ≤3 sentences), 2 follow-ups, positive-reply script, tracking sheet template
7Lead-magnet postSingle post that captures DMsPost only, in a code box. Structure: hook with authority → reframe problem → credibility signal → tease → CTA (connect + comment [trigger] + DM delivery). Strict ban-list enforced.

How it compares to the other LinkedIn-prompt collections

AxisThis thread (7 prompts)Edmondson (5)Godin (6)
Funnel coverageProfile + content + outbound + DMsNarrative + content + proof + commentsPersonal brand (Symbol/Slogan/Surprise/Salient/Story)
SyntaxConversational prose + batched interviewMarkdown headers (#ROLE: / #TASK: etc.)XML tags (<role>/<task>/<steps>/<rules>/<output>)
Specificity of ban-listsHigh — prompt 7 bans named phrasesMedium — bans tips posts, motivational, humble-bragsLow
Interview-first6 of 75 of 56 of 6
Outputs discrete deliverables per promptYesYesYes
Best fitFounder shipping full lead-gen engineExpert establishing thought-leadership angleFounder building personal brand from scratch

All three converge on the “interview first, then output a concrete deliverable” shape — the 2026 standard for applied-prompt threads. Differences are in funnel coverage and the syntax affordance the author happens to prefer.

Implementation

  • Tool/Service: Claude (claude.ai, API, or Claude Code). Prompts are plain prose; run them as-is, paste answers in batches.
  • Setup:
    1. Run prompt 1 first — gives you the profile before any of the content prompts reference it.
    2. Run prompt 2 (Reddit-based) before prompts 3–5 so the content prompts inherit real prospect language.
    3. Prompts 3, 4, 5 all produce posts — avoid running all three in the same week; rotate to keep variety.
    4. Prompt 6 (outbound) is independent of 3–5 but benefits from prompt 2’s pain-point taxonomy and prompt 4’s objection/solution pairs.
    5. Prompt 7 (lead-magnet post) is the “cash prompt” — use whenever a lead magnet launches. Reruns cheaply because the interview is inline.
  • Cost: Free prompts. Running against claude.ai is included in the plan. Running against the API bills per token.
  • Integration notes:
    • Reddit access. Prompt 2 searches Reddit via psychographic queries. In a Claude chat without browsing, paste Reddit results manually. In Claude Cowork or Routines with web access, run queries in-session.
    • Sales-call transcripts. Prompt 4 requires pasted transcripts labeled “call 1 / call 2 / …“. Any call-recording tool (Fathom, Fireflies, Gong, tl;dv, Otter) exports transcripts — paste them directly.
    • Sales Navigator. Prompt 6 assumes Sales Navigator for list building. For founders without SN, substitute LinkedIn basic search or a third-party tool like Apollo / Clay / Instantly.
    • Lead-magnet trigger word. Prompt 7 asks for a “trigger word for DMs” — pick one short, unusual word per lead magnet (this thread used “SYSTEM”). Avoid collisions with common words.
    • WAR headline framework. Net-new vocabulary in this topic — cross-reference from any future LinkedIn article.

Open Questions

  • Attribution. No author or URL supplied. The “WAR framework” and other named constructs are unverified — cite this thread rather than propagating them as general best practice until sourced.
  • Prompt-authorship provenance. This genre (numbered interview-first prompts, X thread format, lead-magnet CTA at the end) is becoming a pattern — many anonymous authors ship near-identical structures. Worth a meta-article on “X-thread prompt genres” once 3+ samples are collected.
  • Efficacy data. Thread claims “100s of people” converted LinkedIn into #1 lead gen. No specific case studies or numbers. Treat as aspirational. ^[inferred]
  • Conflict with Anthropic’s official prompting guidance. Anthropic recommends XML-tagged structure for parseability. This thread uses conversational prose. The trade-off (readability for humans pasting prompts vs parseability for downstream agents) isn’t discussed. Claude Prompting Best Practices leans XML; most founder-facing threads don’t.
  • LinkedIn platform changes. Every LinkedIn-specific prompt has a shelf life — LinkedIn’s algorithm and featured-section mechanics change. Profile-optimization guidance in prompt 1 should be re-verified against the current LinkedIn UI before running.
  • Reddit-search recency. Prompt 2’s psychographic Reddit queries return increasingly outdated results as Reddit content ages. Pair with a recency filter when running.
  • Overlap with existing LinkedIn prompts in this topic. Prompts 3 (weekly content system) and Edmondson’s Prompt 3 (6-week authority calendar) cover similar ground with different frames. No reconciliation article yet — candidate for a connections piece on “applied prompt libraries for LinkedIn.”

Try It

  1. Run prompt 1 today. Even without any other prompts, the profile output is directly actionable — WAR headlines + featured section order + link CTA + banner direction.
  2. Pair prompt 2 with a real Reddit session. Paste 10–20 top Reddit comments from subs where your ICP hangs out. The output is only as good as the raw material; Claude will not surface better language than what you feed in.
  3. Use prompt 4 against a single call first. One call → 7 objections → 7 posts. Calibrate expectations before feeding Claude 20 transcripts.
  4. Treat prompt 7 as the repeatable asset. Every time you ship a lead magnet, fill the Context: [...] block and rerun. The inline-context design makes it cheap.
  5. Compare ban-lists. Diff prompt 7’s banned phrases against Anti-AI Slop Guide and the banned-patterns list in AI Video & Content. Propagate unique bans where they fit.
  6. Cross-compare LinkedIn prompt threads. Run the same brief through prompt 3 (this thread) and Edmondson’s Prompt 3. Keep whichever output actually sounds like you. Threads are frames; voice is yours.
  7. File the WAR framework if it holds up. If the headlines from prompt 1 outperform current profile copy in a week, WAR earns a canonical entry. Until then, keep it scoped to this article.