Marketing

Turns a validated problem into paying customers through the offer, the message, the spread and the growth system. Synthesized from eleven marketing books, each method credited.

Start here

1. Read the venture context if it exists: discovery/00-CONTEXT.md, discovery/LEDGER.md and discovery/FIELD-NOTES.md (own numbers beat book benchmarks). Else ask at most 3 questions: what you sell and to whom, stage and evidence so far, the decision at hand.
2. Gate before tactics: does a reachable group already want this, and does the product deliver? If not, route to audience-and-permission or /customer-discovery. Copy can't fix a market problem.
3. Route with the table below and read only the matching playbook, or two at most. If two rows fit, pick the one matching the buyer's or user's latest action, not their words. Name other relevant playbooks; don't read them.

Router

Principles

1. Steer desire that already exists. Market beats offer, and offer beats copy [Schwartz] [Hormozi] [Godin].
2. Go narrow first. Win a small group completely before you widen [Godin] [Hormozi] [Berger].
3. Marketing amplifies the product and can't replace it. Plug leaks before you pour in traffic [Ellis] [Weinberg] [Eyal].
4. Value is perceived. Try the cheap psychological fix before the expensive logical one, then test it [Sutherland] [Hormozi].
5. Price is a signal. Compete on proof, speed and certainty, not discounts [Hormozi] [Godin] [Sutherland] [Cialdini].
6. Status and identity drive choice. Ask what people like your buyer do, and how buying changes how they are seen [Godin] [Berger] [Heath].
7. Every lever must be true: real limits, real deadlines, real proof. Fake cues borrow against trust [Cialdini] [Hormozi].
8. Remove effort before the first reward. Ask for effort after it [Eyal] [Ellis] [Sutherland].
9. Believe behaviour over opinion. Money, usage and clicks outrank what people say [Sutherland] [Ellis] [Weinberg].
10. Every channel and claim wears out, so keep a pipeline of tests [Weinberg] [Schwartz].
11. Earn before you ask: be useful where your buyers gather before asking them for anything [Godin] [Cialdini] [Weinberg].

When the experts disagree

Output contract

Every answer ends with:

• Recommendation: the call, specific to their business.
• Deliverable: the artifact, filled in (offer stack, headline set, channel ring, test card, hook map).
• Why: the framework(s) behind it, credited [Author].
• Assumptions to log: each one as a ledger item (A#, [ASSUMPTION, E0]) with the cheapest test and its evidence bar. Hand off to /experiments or /customer-discovery. Grades: E0 opinion · E1 told story of past behaviour · E2 observed behaviour/artifact · E3 commitment (time, intro, next step) · E4 money.
• Next action: one thing to do this week.

Research

Web-search only for current, external facts: competitor offers and ads, CPMs and CPCs, platform ad rules, consumer and subscription law, market data. Cite as [FACT — url, date]. Never web-search for "what the books say"; the playbooks are the method. No source → [UNKNOWN] plus a research step. The books are thin on SEO, modern paid social, AI search, B2B demand generation and India-specific distribution (vernacular, WhatsApp, cash on delivery): say so and research rather than guess.

Hand-offs

• Before: /problem-discovery, /customer-discovery, /jtbd, /where-to-play, /value-proposition-canvas.
• Elsewhere: positioning, price and PLG /gtm; names and brand messaging /branding; leads to closed deals /sales; running a push /outreach-sprint.
• After: /experiments, /facts-assumptions-unknowns, /business-model-canvas, /startup-scorecard. Lost: /founder-stack. Learning: /learn.

Go deeper (optional)

If the business-books MCP is connected, search_library / read_passage give the original passages behind a playbook. Not required.