Marketing And Growth

Marketing and Growth Engineering

Marketing work should be driven by business truth + customer evidence + current channel data + measurement, not by a pile of clever prompts.

Use deep references JIT, not all at once:

• broad/substantial marketing strategy, research, positioning, copy, email, content systems -> references/marketing-operating-system.md;
• SEO/search/growth-loop work -> references/seo-growth-loop.md;
• product/feature launches, launch research, video/content production, repurposing -> references/launch-content-system.md.

The first reference distills the useful procedures behind the bookmarked “500 prompts / replace an agency” article into reusable capabilities and workflow chains instead of preserving hundreds of prompts verbatim. The other two distill the Grok SEO/Helena and Okara/Motion/faceless-video bookmark clusters respectively.

Load only the reference(s) that actually apply to the current task.

Core principles

1. Context before copy. Inspect the offer, audience, proof, objections, brand voice, goal, and available evidence before generating campaign assets.
2. Clarity over buzzwords. Communicate a specific truthful value proposition. Avoid generic “revolutionary/seamless/next-generation” filler.
3. Research before positioning. Use customer language, category/competitor evidence, existing analytics, search data, support/review material, and project truth where available.
4. Procedures over magic prompts. Reusable knowledge should be expressed as inputs -> decision/artifact -> evidence/QA -> next stage.
5. Current sources for changing facts. Search behavior, competitors, platform rules, prices/features, APIs, SEO guidance, regulations, and channel norms can change; verify them live when material.
6. Proof before claims. Never fabricate metrics, testimonials, logos, scarcity, guarantees, customer results, or product capabilities.
7. Instrument important funnels. Define the success signal and review window before declaring a campaign/change successful.
8. Close the loop. Baseline -> hypothesis -> action -> measurement -> learning -> keep/iterate/revert/stop.

Capability selection

Load/use only the capabilities the task actually needs. Typical capabilities are:

• customer/ICP research;
• competitor/category analysis;
• positioning and message house;
• content strategy;
• sales/landing copy;
• repurposing/content atomization;
• SEO opportunity + content briefs;
• email lifecycle;
• campaign/launch architecture;
• analytics/CRO/learning loops.

Do not run all ten because a project contains the word “marketing”.

Technical SEO foundation

For implementation-level SEO, verify against the current stack/search-engine guidance and project requirements. Common concerns include:

• crawlability/indexability;
• canonical URLs;
• sitemap/robots behavior;
• semantic page structure;
• structured data appropriate to the actual entity/content;
• metadata/social cards;
• internal linking;
• performance/Core Web Vitals where relevant;
• duplicate/thin/generated content risk;
• locale/hreflang where the site actually needs it.

Do not add every schema type or metadata field mechanically. Match the page and current official guidance.

Copy / value proposition

A useful copy sequence is:

reader state -> desired action -> promise -> mechanism/difference -> proof -> objections -> information sequence -> copy -> factual/brand review

Headlines should explore different strategic angles, not just synonyms. CTA text should make the next action and expectation clear.

Use real social proof only. If no proof exists, change the composition/message rather than manufacturing it.

CRO / funnel optimization

Start from observed or testable friction.

funnel evidence
-> friction hypothesis
-> bounded message/design/product change
-> QA
-> measurement window
-> result
-> learning

Track events that correspond to the real funnel rather than blindly installing a generic event taxonomy.

Content / distribution

A content system should connect audience need to a business objective:

questions / pains / search demand / product evidence
-> content pillars or topic clusters
-> strong source asset
-> platform-native derivatives
-> distribution
-> downstream metrics
-> future topic/format updates

Repurposing may create X/LinkedIn/short-video/newsletter/carousel/etc. derivatives, but it must preserve factual consistency and adapt format/angle to the platform rather than cloning text everywhere.

Campaign / launch

For a material launch, prefer a staged pipeline:

research -> positioning -> message house -> launch brief -> channel architecture -> core assets -> derivatives -> distribution -> questions/data -> follow-up -> review

Use current launch references when they materially improve the decision; do not copy another launch's wording/brand assets or infer causality from a viral post.

Email

Design email sequences from the customer state and job of each message, not from a fixed “5-email template”. Respect the actual jurisdiction/platform's consent, suppression, unsubscribe, and messaging requirements.

Analytics and learning

Choose metrics that match the objective: qualified leads, conversion, revenue/margin, clicks, watch/completion, search performance, retention, objections, etc.

A learning should record evidence, scope, confidence, decision impact, and when to re-check. Do not turn one campaign or two posts into a universal law.

References and provenance

When external sources materially affect a durable marketing decision:

• distinguish canonical/current evidence from creator/vendor claims and inspiration;
• record the actual principle or factual premise used;
• re-verify dynamic facts when needed;
• update the project's reference ledger when the influence is durable.

Large prompt/source collections are treated as source material for procedures, not runtime dependencies. Agency-replacement/pricing framing is not treated as verified outcome evidence.

Completion questions

Before claiming substantial marketing work complete, be able to answer:

• What audience/business problem did this solve?
• What evidence or current sources informed it?
• Which intermediate decisions/artifacts led to the output?
• Are all material claims supportable?
• What metric/evidence will tell us whether it worked?
• What is the review/next-action condition?