SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsx-content-archetypes

x-content-archetypes

Writing & research

View repository

Updated 10-15 23:44

About this skill

Four Content Archetypes — Classified by Reader Action

R — Original Text (Reading)

Don't look at topic tags; look at what readers can do with it. Resource entry type (51%, median engagement 2965), tool tutorial type (39%, median 2035), AI tool discovery type (24%, median 94), ordinary expression type (9%, median 16). A hit isn't about the topic, it's about the function. Classifying by "what readers can do after getting it" shows differences more clearly than topic.

— Xiangyang Qiaomu @vista8, X Viral Secrets Sharing · Four Content Archetypes

I — Methodology Skeleton (Interpretation)

This framework divides content into four categories, not by topic (AI/startups/programming) but by "what readers can do after getting it":

  1. Resource entry type — Readers gain entries that are usually hard to find (tools/documents/websites), median engagement 2965, probability of reaching the top 10% about 40%. Reader action: click to enter, save for later use.
  2. Tool tutorial type — Readers learn to use a certain tool (with steps and a path), median engagement 2035, probability about 24%. Reader action: follow along, save to review later.
  3. AI tool discovery type — Readers see a new tool demonstrated (capability translated into tasks), median engagement 94, probability about 16%. Reader action: understand the possibilities, save to try later.
  4. Ordinary expression type — Readers only read opinions/exclamations/stories, median engagement 16, probability about 9%. Reader action: like (or scroll away).

Core insight:The archetype determines the viral ceiling. The median engagement of the resource entry type is 185 times that of the ordinary expression type. If your content is overly concentrated in the ordinary expression type (opinions/exclamations), the viral ceiling will be locked at a low level. How to use: audit your content structure—what is the proportion of each type? Is the ordinary expression type too much? Convert exclamations into resource entries or tool tutorials.

A1 — Application in the Book (Past Application)

Case 1: Feishu museum document post (resource entry type)

  • Problem: How to make content that "saves readers the search" achieve high spread
  • Use of the methodology: Xiangyang Qiaomu classified this post as resource entry type—providing entries that are usually hard to find (1.55 million items from museums worldwide organized into a Feishu document, directly usable), with the entry placed in the comments
  • Conclusion: The resource entry type has the highest median engagement among the four archetypes (2965), with about a 40% probability of reaching the top 10%
  • Result: 263,000 views, a representative case of the resource entry type

Case 2: ClaudeCode beginner guide post (tool tutorial type)

  • Problem: How to make a tool tutorial lower the threshold for readers
  • Use of the methodology: Classified as tool tutorial type—"even people who don't know how to program can understand it" lowers the trial-and-error cost, with a clear tool name (ClaudeCode) + a complete step-by-step path, and the entry placed in the comments
  • Conclusion: The tool tutorial type has a median engagement of 2035, the second-highest-spread archetype; the key lies in "writing from a beginner's perspective" to lower the threshold
  • Result: 214,000 views, 254 reposts, 1319 saves

Case 3: NotebookLM Skill post (AI tool discovery type)

  • Problem: How to translate abstract AI capabilities into observable results
  • Use of the methodology: Classified as the AI tool discovery type—demonstrating the specific capability of "generating a mind map/podcast/PPT from one sentence," translating tool capabilities into observable task results
  • Conclusion: The AI tool discovery type has a median engagement of 94, lower than the previous two types, but it gains high exposure for a single post through the "discovery narrative" (Damn, this one is even more awesome)
  • Results: 286,000 views, 1,399 saves

A2 — Trigger Scenario (Future Trigger) ★

In what situations would users need this skill?

  1. The user audits their own content structure, finds engagement is low, and wants to know whether it is a problem with the content type
  2. The user posted pure exclamations/opinions (such as "AI is too powerful") and got no response, and wants to know why
  3. The user is planning a content strategy and wants to know which type has the highest reach ceiling and which direction to focus on
  4. The user is judging the reach potential of a post about to be published and wants to know which archetype it belongs to

Language signals (these should activate it when they appear in the user's words)

  • "Why is no one engaging/why is no one responding/why no engagement"
  • "Which content goes viral easily/which content spreads well/which content goes viral"
  • "Is there a problem with the direction of my content/content direction wrong"
  • "Content archetype/content archetype"
  • "Reach ceiling/viral ceiling"
  • "Resource entry type/tool tutorial type/AI tool discovery type/ordinary expression type"
  • "My content structure/content proportion"

Distinction from adjacent skills

  • Difference from x-short-content-craft : This skill does classification and prediction (which archetype your content belongs to, how high its reach ceiling is), while the latter does production (how to write Hook-Body-CTA after the type has been chosen).Classify first, then produce。
  • Difference from x-five-piece-checklist : This skill looks at "archetype type" to determine the reachceiling, while the latter looks at "element completeness" to determine a single post'sfloor. If the archetype is wrong, even a complete five-piece set cannot raise it.
  • Difference from x-four-saves : This skill predicts reach by dividing content into four categories based on "what readers can do," while the four-savings model evaluates value by "how many steps it saves readers." They can be combined: first determine the archetype, then verify the four savings.
  • Difference from x-longtail-strategy : This skill provides a classification framework to choose high-probability archetypes, while the long-tail strategy provides a mindset framework to accept the median as normal.
  • Difference from x-positioning-tradeoff : This skill focuses on content classification and reach prediction, while the latter focuses on account positioning trade-offs (IP vs. broad traffic).

E — Executable Steps (Execution)

After the skill is activated, the agent should execute the following steps:

  1. Audit the existing content structure

    • List the user's recent posts (10-20 recommended), and classify each one into one of the four archetypes
    • Classification criteria: What can readers do after receiving it? Has an entry point → resource entry type; has steps → tool tutorial type; has a tool demonstration → AI tool discovery type; only opinions/exclamations → ordinary expression type
    • Completion standard: Each post is labeled with its archetype type, and the proportion of each type is counted

  2. Diagnose reach bottlenecks

    • Compare against the median engagement data for each archetype (resource entry 2965 / tool tutorial 2035 / AI discovery 94 / ordinary expression 16)
    • Identify archetypes with an excessively high proportion but low median engagement (usually too many ordinary expression type posts)
    • Completion criteria: Clearly state "your X% of content is concentrated in plain expression type, with a median engagement of only 16, which is the distribution bottleneck"

  3. Low-conversion prototype

    • Convert plain expression type into resource entry type (add entry points: tool/document/website links) or tool tutorial type (add steps: a usage path within three steps)
    • Stop condition: If the user's content itself is opinion type/story type and is intentional (not an accidental exclamation), then do not force conversion; instead, advise keeping the proportion of plain expression type below 20%
    • Completion criteria: At least one original "exclamation/opinion" post is rewritten into a version containing an entry point or steps, or it is explicitly recommended to reduce the proportion of plain expression type

B — Boundary ★

Do not use this skill in the following situations

  • A certain short content type (story/list/opinion/tutorial) has already been determined and the Hook-Body-CTA structure needs to be organized → use x-short-content-craft. This skill is classification prediction, not structure production.
  • Check whether a single post's five-piece set (value promise/scenario/entry point/evidence/reason to save) is complete → use x-five-piece-checklist. This skill looks at the prototype ceiling, while the latter looks at the element floor.
  • Make account positioning trade-offs (IP vs broad traffic) → use x-positioning-tradeoff. This skill does not involve positioning decisions.

Failure modes the author warns about in the book

  • ce13 (broad traffic = too much plain expression type): Accounts pursuing broad traffic often have an excessively high proportion of plain expression type; followers follow for various reasons (entertainment/bandwagon), have no trust relationship with the creator, brands do not pay, and monetization conversion is extremely poor. Content on the broad traffic path naturally leans toward plain expression type, forming a dual dilemma of "low distribution + low monetization."
  • ce15 (overfitting sample): The prototype proportion data (51%/39%/24%/9%) and median engagement data come from Xiangyang Qiaomu's personal 3.4G data (AI/product research track, Chinese-speaking audience, 2023-2026 algorithm versions). Different tracks, different audiences, and different periods have different distributions. Use asdirectional referencerather than hard metrics — resource entry type is very likely superior to plain expression type, but the specific multiple varies by person.

The author's blind spots / era limitations

  • The data comes from a single account (Xiangyang Qiaomu), with no control group, and cannot prove that "prototype → distribution" is a stable causal relationship rather than correlation
  • The median engagement data is point-in-time data; after X algorithm updates (such as changes in OON proportion, removal of link downranking), the performance of each prototype may drift
  • The four prototypes are the author's personal summary and are not an exhaustive classification — some content (such as pure entertainment/jokes) may not fit within this framework
  • The perspective of "classifying by reader action" itself is transferable, but the specific division of the four categories and the values will evolve with the platform ecosystem

Adjacent methodologies that are easy to confuse

  • Four short content types(story/list/opinion/tutorial) classify by "content form," while the four content prototypes classify by "reader action"; the two classification systems are orthogonal — a post can simultaneously be "list type (form) + resource entry type (action)"
  • Four Provinces ModelEvaluates the value of a single item by "saving readers a few steps," while the four prototypes predict the distribution ceiling by "what readers can do"; the former is a valuation tool and the latter is a classification tool

Related skills

  • depends-on x-short-content-craft: Before drafting short content, first understand the distribution differences among the four prototypes; the prototype determines which short content type should be used for writing — classify first, then produce.
  • contrasts-with x-comment-engagement: Content leverage for gaining followers (choose high-distribution prototypes so posts run on their own) vs engagement leverage for gaining followers (seize the comment section + Quote to borrow others' traffic); see the stage and where to invest effort.

Audit Information

  • Verification Passed: V1 ✓ / V2 ✓ / V3 ✓
  • Test Pass Rate: Pending test (see test-prompts.json for details)
  • Distillation Time: 2026-07-14