Marketing Strategy Master

Marketing Strategy Master

Implements: commons:strategy-master-pattern — 9-section living document structure, diff-only write rules, append-only changelog. This skill configures the pattern for the marketing domain (H-### hypotheses, channel-by-channel health, qualified leads north star).

Owner of the single living strategy doc. Every other marketing-strategy skill reads this doc for current priorities; only this skill writes to it.

Dependencies

Load marketing-identity first. It defines ICP, hook, and hypothesis format.

Also read:

• /var/www/html/systemprompt-web/reports/sales-marketing-strategy.md — master sales & marketing strategy. Channel priorities in this doc's §7 must align with the master strategy's §4 domain weights. When updating channel priorities, cross-reference the master strategy.

The Strategy Doc

/var/www/html/systemprompt-web/reports/marketing/marketing-strategy-master.md

Lives inside reports/marketing/ alongside the dated lead-tracker.md / daily-brief.md outputs and the drafts/ folder. Single living file, not dated — diffs are appended, never rewritten.

Structure (enforced on every write):

# systemprompt.io — Marketing Strategy Master

**Last updated:** {YYYY-MM-DD by {skill-name}}
**Phase:** {1: Baseline | 2: Test Channels | 3: Double Down | 4: Scale}
**North Star:** Qualified Leads/week (lead = template-ran + feedback)

## 1. Current Funnel Snapshot
(tables updated by lead-tracker; this skill only reads, never overwrites)

## 2. Objectives — 30 / 60 / 90
| Horizon | Template unique cloners/week | Feedback leads | Qualified convos | Sandbox | Ref customers |
|---|---|---|---|---|---|
| 30d | {N} | {N} | {N} | {N} | {N} |
| 60d | {N} | {N} | {N} | {N} | {N} |
| 90d | {N} | {N} | {N} | {N} | {N} |

## 3. Active Hypotheses (in flight)
Table of hypotheses currently being tested — owned by `hypothesis-ledger`, rendered here.

## 4. Winning Tactics (validated)
Tactics where hypothesis tests passed. Short paragraph per tactic: what, why, playbook.

## 5. Dead Hypotheses (and why)
Short table: hypothesis, result, reason for failure, lesson.

## 6. This Week's Theme
One sentence theme, 3–5 message pillars, list of draft hooks.

## 7. Channel Priorities (ranked by leads/Ed-hour)
Ranked list with current weight (% of Ed's outreach hours).

## 8. Instrumentation Gaps
Things we cannot yet measure. Listed so we remember to fix them.

## 9. Changelog
Append-only. One line per diff, newest first: `{YYYY-MM-DD} {skill-name} — {what changed}`

Write Rules

This skill NEVER rewrites the doc wholesale. It only applies targeted diffs, and every diff appends a line to §9 Changelog.

Permitted operations:

Sections §1 (funnel snapshot) and §3 (active hypotheses) are rendered — the actual source is lead-tracker output and hypothesis-ledger. This skill only re-renders them, never stores derived data.

Run Modes

marketing-strategy-master seed        # First-run: create the doc if missing
marketing-strategy-master read        # Print the doc (default for other skills)
marketing-strategy-master diff {...}  # Apply a targeted diff + changelog entry
marketing-strategy-master render      # Re-render §1 and §3 from lead-tracker + hypothesis-ledger

Seed Data (first run)

Use real baseline captured on 2026-04-15:

• Template repo (14d): 81 views, 11 unique viewers, 1,134 clones, 162 unique cloners (Apr 14 spike of 959/119 flagged as CI bots — treat unique cloners as honest signal). Referrers: github.com, systemprompt.io.
• Core repo (14d): 112 views, 8 unique viewers, 128 clones, 47 unique cloners. Referrers: github.com, crates.io, docs.rs, systemprompt.io.
• Website (7d): 165 sessions, 107 unique users, 389s avg duration, 2.28 requests/session. Top pages: claude-skills-non-technical-teams, mcp-vs-cli-tools, claude-plugins-vs-mcp-vs-skills. Traffic sources include GitHub (~30%), ProductHunt, Bing, Slack.
• Stars: 1 each (both self-stars).
• External feedback leads: 0.

30/60/90 targets (starter — to be recalibrated after Phase 1 baseline week):

Instrumentation Gaps (known at seed time)

• No GitHub Discussions enabled on either repo → feedback path routes through Issues with feedback label, plus hello@systemprompt.io. Needs a decision: enable Discussions?
• GSC service account key lives at /var/www/html/systemprompt-web/.keys/gsc.json (NOT in the marketplace repo). Working as of 2026-04-15; service account gsc-559@gen-lang-client-0891438583.iam.gserviceaccount.com on site sc-domain:systemprompt.io.
• No way to deduplicate "CI bot clones" from "human clones" beyond the unique-cloner count. Large raw-clone spikes get flagged but not removed.
• LinkedIn/X impressions are manual paste-in only (no API access yet).

Anti-Sludge Rules

• Every diff must reference the hypothesis ID or data source that motivated it.
• Numbers over vibes: "template unique cloners/wk rose from 30 to 72" not "clones growing."
• No sections added without a hypothesis attached.
• If a diff would make the doc contradict itself (e.g., channel ranked #1 in §7 but all its hypotheses in §5), abort and surface the conflict to Ed.