Rama Marketing
An open, evidence-led agent skill for paid-media analysis and strategy across Meta Ads, Google Ads, TikTok Ads, and LinkedIn Ads.
Rama Marketing turns campaign exports, business economics, complete customer-journey evidence, and current platform documentation into an auditable diagnosis, experiment plan, and prioritized 30/60/90-day strategy. It does not promise automatic growth, fabricate benchmarks, or treat platform attribution as causal proof.
The research library contains 14 routed references, more than 120,000 words, and more than 560 unique direct source URLs.
Research baseline: 2026-07-21. Volatile platform facts must be rechecked against their linked official source at use time.
Install
Install the skill into compatible project-local agent roots with the open skills CLI:
npx skills add RamaAditya49/rama-marketing \
--skill rama-marketing \
-a claude-code -a codex -a openclaw -a pi -a antigravity \
--copy -y
For compatible global roots (Codex global uses the native plugin route below):
npx skills add RamaAditya49/rama-marketing \
--skill rama-marketing \
-a claude-code -a hermes-agent -a openclaw -a pi \
-g --copy -y
List the skill before installing:
npx skills add RamaAditya49/rama-marketing --list
The common SKILL.md package is discoverable by Codex/Codex CLI, Claude Code, Hermes Agent, OpenClaw, Pi, Antigravity, and other compatible agents. For the native Codex/ChatGPT plugin package:
codex plugin marketplace add RamaAditya49/rama-marketing --ref main
codex plugin add rama-marketing@rama-marketing
Start a new session after installation. ChatGPT plugin availability is surface-dependent: current official documentation supports plugins in Work mode on the web and in Work mode or Codex in the desktop app, but not Chat mode, the IDE extension, or mobile. See agent compatibility for exact platform limits and non-CLI routes.
Claude Code also has a native marketplace package. Run these inside Claude Code:
/plugin marketplace add RamaAditya49/rama-marketing
/plugin install rama-marketing@rama-marketing
Ask it
Use $rama-marketing to audit these Meta, Google, TikTok, and LinkedIn exports. Reconcile them with CRM revenue, identify the binding constraint, and return a 30/60/90-day plan.
Use $rama-marketing to build a paid acquisition strategy for a B2B SaaS product with a 90-day sales cycle, $25k ACV, and limited creative capacity. Show assumptions, channel scorecard, experiments, and stop/scale rules.
Use $rama-marketing to determine whether our reported ROAS decline is tracking, auction, creative, landing-page, sales-quality, or market related. State what is observed, calculated, verified, and still hypothetical.
Use $rama-marketing to map our complete AOF–TOF–MOF–BOF–activation–DOF–advocacy journey. Separate observed customer states from campaign labels and identify the highest-value transition to improve.
Use $rama-marketing to audit Ad → landing page → WhatsApp click → real inbound chat → response → qualified → hot → closed-won. Reconcile IDs, calculate every stage rate, diagnose spam, and recommend lawful Meta feedback signals.
Use $rama-marketing to design our audience system across Meta, Google, TikTok, and LinkedIn. Separate strategic segments from platform audiences; audit Customer Match, Custom/Matched, Lookalike/Predictive, signals, expansion, exclusions, seed quality, journey evidence, creative jobs, and incrementality.
What is inside
- Current platform research with direct official-source registers and volatility notes.
- Cross-platform channel, objective, bidding, maturity, and budget matrices.
- Unit-economics, attribution, incrementality, experiment, and uncertainty methods.
- Ethical customer research and behavioral-science guardrails.
- A cross-platform audience ontology and decision system for strategic segments, first-party cohorts, Custom/Matched/Customer Match audiences, Lookalike/Predictive models, signals, controls, expansion, and actual delivery.
- A nonlinear AOF–TOF–MOF–BOF–DOF model with separate demand, value, and advocacy states.
- An ad-to-landing-to-WhatsApp/CRM/closing measurement and optimization playbook.
- Vertical playbooks for ecommerce, B2B/SaaS, local services, apps, education, property, automotive, healthcare, finance, nonprofits, events, and marketplaces.
- Data contract and a zero-dependency CSV analyzer for deterministic KPI aggregation.
- Governance for privacy, regulated categories, discrimination, fraud, brand safety, and source quality.
The skill routes references progressively so an agent loads only what the decision requires.
The research library is bilingual (English and Indonesian); the skill always answers in the user's language.
Analyze an export
Normalize a CSV to the data contract, then run:
python3 skills/rama-marketing/scripts/analyze_paid_media.py campaigns.csv > audit.md
For JSON or another breakdown:
python3 skills/rama-marketing/scripts/analyze_paid_media.py campaigns.csv --format json --group-by platform,campaign
The analyzer uses aggregate numerators and denominators, calculates optional WhatsApp/chat-to-qualified-to-close metrics, tracks structural-null coverage, rejects incomparable context values, preserves decimal precision in JSON, and emits deterministic data-quality flags. Its output is descriptive; it never claims incremental lift.
Evidence standard
Every consequential claim is classified as:
observed: present in supplied or queried data;calculated: derived with a visible formula;verified: supported by a cited external source and access date;hypothesis: plausible but still requires a discriminating test.
Recommendations must connect to contribution margin, allowable acquisition cost, payback, capacity, lead/customer quality, and uncertainty. Universal benchmark tables are intentionally excluded because market, objective, attribution, geography, auction, and business economics make them unsafe defaults.
Library
| Reference | Use it for |
|---|
| Meta Ads | Delivery, Advantage+, signals, measurement, API, audit |
| Google Ads | Search, Shopping, PMax, YouTube, bidding, measurement, audit |
| TikTok Ads | Smart+, Spark, creative, commerce, measurement, audit |
| LinkedIn Ads | B2B targeting, Accelerate, lead gen, CRM/revenue measurement |
| Measurement and experiments | Attribution, lift, power, lag, MMM, causal decisions |
| Customer psychology and ethics | Research, behavior, segmentation, ethical persuasion |
| Audience segmentation and activation | Segments, Custom/Matched/Customer Match, Lookalike/Predictive, signals, exclusions, seed quality, delivery, and experiments |
| Complete customer journey | AOF–DOF, advocacy, WhatsApp/CRM stages, spam feedback, audience overlap |
| Industry playbooks | Business-model constraints and vertical decision patterns |
| Creative, offer, and landing | Insight-to-creative system and post-click diagnosis |
Contribute
Open an issue or pull request with the claim being changed, direct source URL, publication/update date, access date, affected market, and whether the change is a platform fact or a practitioner hypothesis. Do not add unsourced benchmark claims or copied proprietary playbooks.
License
MIT © 2026 Rama Aditya.
