LinkedIn Editorial Policy and Author Voice

Purpose

This document governs how Professional Radar produces LinkedIn draft candidates from site/radar signals. It ensures output remains human-reviewed, low-frequency, source-backed, and clearly useful.

Core Principles

• Draft-only. No automatic publishing.
• No scheduled posting. No automation of any LinkedIn interaction.
• Manual approval required. Every draft must be reviewed before publish.
• Quality over quantity. Max 2-3 strong candidates per week.
• Source-backed. Every claim must have a source URL.
• No confidential data. Never use employer/client-specific information.

Prohibited Actions

The following are explicitly forbidden:

• Automatic publishing to LinkedIn
• Scheduled LinkedIn posting
• Automatic commenting, liking, or reposting
• Automatic connection requests or direct messages
• Generic promotional calls to action
• Fabricated personal experience or client stories
• Invented emotions, achievements, numbers, or results

Quality Gate

Each LinkedIn draft candidate must pass these checks:

1. Human Value

• Specific insight, practical lesson, or decision implication
• Not just a summary of a news item
• Clear relevance to SAP AMS, SD/MM, clean core, data, integration, or AI-enabled delivery

2. Originality

• Not copied from source text
• Includes professional interpretation
• Not near-duplicate of recent drafts or published items

3. Source Backing

• Source URLs in metadata
• No invented claims
• No confidential employer/client data

4. Anti-Repetition

• Compare topic, angle, opening sentence, structure, and final takeaway against recent content
• Reject or rewrite if too similar
• Do not reuse the same opening style twice in a row
• Do not use the same post structure more than once per week

5. Professional Tone

• No hype
• No clickbait
• No excessive hashtags
• No generic AI buzzword posts
• Concise professional English (B2 level)

6. Frequency/Cooldown

• Do not generate another draft for the same topic within 7 days unless explicitly requested
• Default: max 2-3 strong candidates per week
• Prefer weekly digest for multiple weak/moderate signals

Draft Status Model

Required Draft Metadata

Every draft must include:

draft_id: "linkedin-2026-05-27-sap-ai-units"
source_signal_id: "signal-2026-05-27-001"
source_urls:
  - "https://help.sap.com/docs/SAP_S4HANA_CLOUD"
topic: "sap_business_ai"
angle: "cost_implication"
target_audience: "sap_ams_leads"
why_it_matters: "AI Units pricing changes affect AMS budget planning"
duplicate_check_result: "no_similar_recent"
quality_risk: low  # low | medium | high
recommended_publish_window: "2026-05-29"
status: drafted
style_pattern: consultant_takeaway
opening_type: problem_first
repeated_structure_check: pass
similar_recent_post: null
human_voice_score: 4  # 1-5
template_risk: low  # low | medium | high
rewrite_reason: null

Rejection Rules

Reject draft generation if:

• Source is weak or unclear
• Topic is too generic
• Draft is mostly summary without professional insight
• Draft is too similar to recent content
• Content sounds promotional without substance
• It needs confidential/client-specific examples
• It would create too many posts in a short period

Author Voice

Preferred Voice

• Direct
• Practical
• Slightly skeptical
• Operational, not inspirational
• Clear B2-level English
• Short paragraphs
• Concrete SAP/AMS/process implications
• One useful idea per post
• No hype

Banned Patterns

Never use:

• “In today’s fast-paced world…”
• “AI is transforming everything…”
• “Here are 5 reasons why…”
• “I’m excited to share…”
• “Game-changer”
• “Revolutionary”
• “Unlock the power of…”
• “The future of X is here”
• “This got me thinking…”
• Repeated emoji bullets
• Hook → 3 bullets → CTA template
• Generic “What do you think?” ending
• Excessive hashtags

Style Patterns

Each draft must intentionally choose one pattern:

1. signal_note

Start with the external signal. Explain why it matters. End with operational implication.

2. consultant_takeaway

Start with a practical SAP/AMS problem. Connect source signal to that problem. End with what teams should check.

3. risk_note

Start with a risk or hidden cost. Explain the mechanism. End with a small control/check.

4. process_note

Start with a process failure mode. Explain where AI/automation helps or does not help. End with a realistic next step.

5. contrarian_note

Start with what people may overestimate. Explain what matters more. End with balanced takeaway.

6. mini_field_checklist

Start with a short context line. Provide 3-5 checks. End without weak engagement bait.

7. weekly_digest

Use only when there are multiple weak/moderate signals. Group into one useful summary. Do not create separate posts for every small signal.

Anti-Repetition Rules

Before creating a draft:

• Compare with recent LinkedIn drafts and published records
• Check topic, angle, opening sentence, structure, and final takeaway
• Reject or rewrite if too similar
• Do not reuse the same opening style twice in a row
• Do not use the same post structure more than once per week
• Do not repeat the same CTA style

Rewrite Process

If a draft feels generic:

1. Remove hype
2. Remove generic opening
3. Add one concrete SAP/AMS/process implication
4. Shorten
5. Change structure
6. Remove weak CTA
7. Re-check similarity

Two-Variant Generation (Optional)

For each approved signal, optionally generate exactly two draft variants:

• Variant A: practical consultant note
• Variant B: risk/process takeaway

Then select the stronger one and mark the other as rejected_variant. Do not save both as active drafts unless explicitly requested.

Quality Gate Checklist

A draft is acceptable only if:

• human_voice_score >= 4
• template_risk = low
• repeated_structure_check = pass
• Has a specific SAP/AMS/process takeaway
• Does not sound like generic marketing content

Sample Accepted Draft

draft_id: "linkedin-2026-05-20-clean-core-dashboard"
source_signal_id: "signal-2026-05-15-002"
source_urls:
  - "https://support.sap.com/en/alm/sap-cloud-alm.html"
topic: "sap_clean_core"
angle: "operational_visibility"
target_audience: "sap_ams_leads"
why_it_matters: "Clean Core dashboard gives AMS teams visibility into custom code health without manual audits"
duplicate_check_result: "no_similar_recent"
quality_risk: low
recommended_publish_window: "2026-05-22"
status: approved_manually
style_pattern: consultant_takeaway
opening_type: problem_first
repeated_structure_check: pass
similar_recent_post: null
human_voice_score: 4
template_risk: low

Draft body:

Most AMS teams discover custom code issues during upgrades or incidents. By then, the cleanup cost is already high.

SAP Cloud ALM now has a Clean Core dashboard that shows custom code health continuously. For AMS teams, this means proactive monitoring instead of reactive cleanup.

What to check:

• Is your custom code inventory mapped against SAP standard objects?
• Are you tracking compatibility pack deadlines?
• Do you have a plan for custom code remediation before the next upgrade cycle?

The dashboard won’t replace architecture discipline. But it gives AMS teams a starting point for conversations about technical debt that were previously hard to quantify.

Sample Rejected Draft

draft_id: "linkedin-2026-05-20-ai-revolution"
source_signal_id: "signal-2026-05-20-003"
source_urls:
  - "https://news.sap.com/ai-announcement"
topic: "sap_business_ai"
angle: "generic_hype"
target_audience: "general"
why_it_matters: null
duplicate_check_result: "similar_to_draft-2026-05-15"
quality_risk: high
recommended_publish_window: null
status: rejected
style_pattern: signal_note
opening_type: generic_hype
repeated_structure_check: fail
similar_recent_post: "draft-2026-05-15"
human_voice_score: 2
template_risk: high
rewrite_reason: "Generic hype opening, no operational takeaway, similar to recent post"

Rejection reason: Generic hype language (“revolutionary”, “game-changer”), no specific SAP/AMS implication, too similar to recent post about AI adoption.

No Automatic Publishing

This policy explicitly prohibits any automatic LinkedIn publishing. All drafts remain in drafted, needs_review, or approved_manually status until a human manually publishes them.

Related

• #18 — LinkedIn editorial quality and manual approval policy
• #19 — LinkedIn author voice and style variation gate
• docs/content-taxonomy.md — Content type definitions
• docs/classification-rules.md — Content classification pipeline