Marketing Compass

View saved version →

---
name: articulate-marketing-problem
description: "Turn scattered marketing or business symptoms, metrics, stakeholder statements, disagreements, and vague concerns into an evidence-bounded, testable problem definition before structural diagnosis or solution design. Use when a user says the issue is unclear, teams describe different problems, observations exist only as disconnected points, a request already jumps to ads/CRM/MA/reorganization, metrics conflict, definitions may have changed, or source data and interviews must be gathered to create an analysis-ready brief. Separate facts, measurements, interpretations, emotions, hypotheses, requests, and proposed solutions; connect points without inventing causality; identify competing explanations; and specify or perform the minimum evidence acquisition that can change the decision."
---

# Articulate Marketing Problem

Create the problem before solving it. Convert fragmented reality into an analysis-ready brief without polishing assumptions into facts or declaring a single root cause prematurely.

Read [references/problem-formation.md](references/problem-formation.md) for every substantive problem-definition task. Read [references/evidence-acquisition.md](references/evidence-acquisition.md) when files, systems, interviews, definitions, or additional data must be inspected or requested. Read [references/output-contract.md](references/output-contract.md) before returning a full brief.

## Establish the decision context

Identify:

- Who must decide what, and by when
- What triggered the concern
- What outcome, customer state, or operating condition is at risk
- What action is already being proposed, by whom, and on what premise
- What evidence is available now and what access is actually authorized

Ask at most three initial questions that materially change the scope or evidence plan. If answers are unavailable, continue with a provisional problem statement and label its limits.

Do not accept the requested solution as the problem. Translate “increase advertising,” “replace CRM,” “sales must follow up,” or “we need more people” into the condition those actions are intended to change.

## Phase A — Collect and classify the points

Extract atomic statements from conversation, documents, dashboards, spreadsheets, meeting notes, interviews, and system exports. Preserve source and wording when material.

Classify every consequential point as one of:

- Confirmed event or directly observed fact
- Measurement, including definition, denominator, population, period, and source
- Stakeholder report or lived observation
- Interpretation or causal story
- Emotion, concern, incentive, or desired outcome
- Hypothesis
- Unknown
- Proposed action or solution

Do not collapse stakeholder reports into system facts. Do not treat a logged value as self-explanatory truth; the value can be real while its definition, population, or comparison is unsuitable.

## Phase B — Connect points into candidate structures

Connect points only through an explicit relationship:

- Same population or cohort
- Temporal sequence
- Stage or process dependency
- Shared constraint or common cause
- Feedback loop or delay
- Contradiction or definition mismatch
- Stakeholder perspectives on the same event

Mark each connection as `confirmed`, `supported`, `plausible`, or `unknown`. Never join different cohorts, periods, definitions, or attribution systems silently.

Generate at least two competing explanations when the decision is consequential. State what observation would distinguish them. If the points cannot yet support a structure, say so and produce an evidence-acquisition plan instead of a polished narrative.

## Phase C — Form the problem statement

Write the narrowest useful problem definition that states:

1. Scope: population, process, market, cohort, and period
2. Observed condition or change
3. Why it matters to the pending decision
4. Candidate mechanism, explicitly labeled by evidence status
5. Material competing explanation
6. Missing evidence that could reverse the definition

Prefer formulations such as:

> In scope S and period T, observations O show condition C. This prevents or threatens decision outcome R. Evidence is consistent with mechanism H1, while H2 remains plausible because evidence E is missing. The next useful step is V.

Do not claim to have found the “true problem” when the evidence supports only a provisional issue or decision uncertainty.

## Phase D — Acquire the minimum decisive evidence

When authorized files, systems, or connectors are available, inspect them rather than merely listing unknowns. Follow [references/evidence-acquisition.md](references/evidence-acquisition.md): verify definitions and lineage first, then retrieve only information that distinguishes competing problem definitions.

When direct access is unavailable, specify:

- Exact field, extract, document, or interview evidence needed
- Source system or responsible person
- Population, period, cohort, and granularity
- Why the evidence changes the decision
- What result supports, weakens, or refutes each explanation
- Owner and practical timing when known

Do not ask for exhaustive discovery. Stop when the remaining uncertainty no longer changes the immediate reversible decision, or when the acquisition cost exceeds the decision value.

## Route without overreaching

Return an analysis-ready problem brief, not a full marketing strategy.

- Route to `diagnose-marketing-structure` after the issue and evidence base are sufficiently defined to compare bottlenecks.
- Route to `audit-marketing-reasoning` when an existing report or proposal itself is the audit object.
- Route to `design-marketing-measurement` when metric architecture, experiments, causal identification, or thresholds must be designed.
- Route to the relevant advertising, B2B, MA/CRM/LTV, or communications skill only after stating the defined problem that justifies it.

If an immediate danger, legal deadline, material customer harm, or irreversible loss exists, separate containment from deeper problem formation and prioritize the safe immediate action.

## Return a usable brief

Adapt depth to the material. For ambiguous but small requests, return a compact articulation. For cross-functional or consequential cases, use the full format in [references/output-contract.md](references/output-contract.md).

Always distinguish:

- What is happening
- What people think it means
- What may connect the points
- What is still missing
- What the problem can defensibly be called now
- What should be checked or collected next

## Guardrails

- Do not turn a complaint, metric, or proposed solution into the problem by paraphrasing it.
- Do not invent a link merely because multiple symptoms form an elegant story.
- Do not merge different cohorts, periods, definitions, or sources without qualification.
- Do not interpret stakeholder disagreement as evidence that one party is incompetent or acting in bad faith.
- Do not gather data because it is available; gather it because it can distinguish decisions or hypotheses.
- Do not let missing information become a reason for endless discovery or no decision.
- Do not expose credentials, private customer data, or unnecessary personal information while gathering evidence.
- Do not diagnose the maximum bottleneck before the problem is sufficiently formed; hand off to structural diagnosis.
- Do not present Marketing Compass terminology as universal industry terminology.

Referenced files: 5

View saved version →

---
name: assess-ma-crm-ltv
description: "Assess whether marketing automation, CRM, lifecycle messaging, customer success, retention programs, or LTV initiatives fit the business and what must be fixed first. Use when a user asks whether to introduce or strengthen MA, why CRM is not increasing revenue, how to separate strategic CRM, operational CRM, MA, and CS, why retention or renewal is weak, how to improve LTV, how to design lifecycle journeys or automation, or how to distinguish accounting LTV, acquisition economics, and the strategic lifetime of customer choice."
---

# Assess MA, CRM, and LTV

Diagnose the relationship and value structure before automating contact. Treat MA as an execution device, CRM as both a relationship strategy and an operating function, CS as value-realization intervention, and LTV as both an accounting result and a strategic formation problem.

Read [references/role-and-readiness.md](references/role-and-readiness.md) for every substantive diagnosis. Read [references/ltv-model.md](references/ltv-model.md) when calculating or diagnosing LTV. Read [references/output-contract.md](references/output-contract.md) before returning a complete recommendation.

## Establish the decision

Identify:

- Who must decide what and by when
- The business model, buying cycle, and repeat or continuation mechanism
- The customer population and cohort in scope
- The current customer state and observed problem
- The proposed MA, CRM, CS, or retention intervention
- Available behavioral, transactional, attitudinal, and service data
- Capacity, permission, customer risk, and acceptable loss

If the request begins with a tool or campaign, restate the underlying decision. Ask at most three questions that can change the readiness decision, then continue provisionally when possible.

## Test critical prerequisites

Check the three major blockers before using a scorecard:

1. Does relevant base or recognized demand exist?
2. Is the ICP or target customer state defined?
3. Does the customer reach and experience the promised value?

If any major blocker is absent, default to `Hold` or `Adverse` for additional automation. Do not let a majority of weaker positive conditions overrule a structural failure.

Then assess:

- Demand repeats or a meaningful future decision recurs
- A reason to choose already exists
- Timing or friction can change the outcome
- Customer states can be observed or responsibly inferred
- Contact permission and channel suitability exist
- The organization can act on signals
- Human intervention is available where ambiguity is high

Treat any `3/5` or `2/5` rule as a rough prompt, not a formal threshold.

## Separate responsibilities

Use these roles:

- **Strategic CRM:** design customer states, relationship meaning, value timing, contact logic, and LTV structure.
- **Operational CRM:** prevent leakage, restore recall, support return, adjust timing, and reduce selection friction.
- **MA:** automate contact, routing, timing, reminders, and known friction.
- **CS:** help customers reach value, succeed, adapt, and form a reason to continue.

Do not reduce CRM to message delivery, CS to support tickets, or MA to a revenue engine. MA does not create base demand, ICP, product value, or competitive advantage.

## Map the customer-state system

Use the smallest relevant path:

```text
Need recurrence
  → recall or return trigger
  → re-selection
  → access and friction
  → use and value realization
  → continuation, repurchase, expansion, or advocacy
```

For subscription or SaaS, include acquisition, activation, first value, habit, renewal, expansion, and recovery where relevant.

For each state, separate:

- Confirmed facts
- Proxy signals
- Hypotheses
- Unknowns
- Customer risk
- Appropriate automated and human response

Select one primary bottleneck. Do not launch a full lifecycle program to solve every stage simultaneously.

## Diagnose value realization and expectation

Check:

- What value the customer expected before purchase
- The earliest moment that value can be experienced
- Time and effort required to reach it
- Product, process, data, integration, or behavioral blockers
- Who owns activation and success
- Whether messaging created an expectation the experience cannot meet
- Whether churn reflects lack of demand, poor fit, poor value, friction, or avoidable forgetting

If value realization is weak, prioritize product, onboarding, implementation, or CS. Do not replace them with reminders or discounts.

## Design contact by purpose

Every contact must have one legitimate role:

- Restore relevant recall
- Help complete an intended action
- Remove known friction
- Support value realization
- Prepare a recurring need or decision
- Surface risk for human intervention
- Confirm preference or permission

Define the entry condition, customer benefit, timing, channel, suppression rule, exit condition, and owner.

Do not contact because a tool allows it. Use frequency, complaint, opt-out, negative engagement, trust, and value signals as guardrails.

## Separate automation and human judgment

Automate high-volume, repeatable, low-ambiguity actions. Use people for:

- Unclear needs or account context
- Failure to reach value
- Sensitive recovery or dissatisfaction
- Complex tradeoffs and relationship risk
- New objections or exceptional behavior
- High-value intervention where consent and judgment matter

Define escalation and suppression rules. Do not simulate a human relationship through excessive automated personalization.

## Diagnose LTV correctly

Use three separate views:

1. **Accounting LTV:** period gross-margin cash contribution less customer maintenance and service costs.
2. **Acquisition economics:** accounting LTV compared with CAC and payback.
3. **Strategic LTV:** the durability of being chosen again—the lifetime of choice.

Use [references/ltv-model.md](references/ltv-model.md) for definitions and calculation limits. Never calculate the strategic formation tree as if its factors were calibrated probabilities.

Diagnose strategic LTV through:

- Probability that the relevant need recurs
- Probability of being recalled again
- Probability of being re-selected
- Probability of continued use or repurchase after value realization
- Contribution margin per decision or period

State cohort, period, margin basis, service cost, churn convention, and discounting when making financial claims.

## Decide MA and relationship action

Choose one:

- Adopt or expand
- Conditional pilot
- Hold until prerequisites
- Reduce or redesign
- Adverse / likely harmful

Specify:

```text
Desired Signal
Counter-signal
Guardrail
Review window
Action if crossed
```

Prefer a narrow use case that tests the relationship hypothesis over platform-wide automation.

## Return a diagnosis

Use the compact or full structure in [references/output-contract.md](references/output-contract.md). Include:

- MA readiness and major blockers
- Strategic CRM, operational CRM, MA, and CS responsibility
- Primary customer-state bottleneck
- LTV view and formation drivers
- What to fix before automation
- Automated action and human-intervention point
- Signal, counter-signal, guardrail, review window, and action
- What not to do

## Guardrails

- Do not use MA to replace demand, ICP, value, or competitive advantage.
- Do not treat more messages as stronger customer relationships.
- Do not call opens, clicks, or recency customer value or loyalty.
- Do not automate contact without purpose, permission, suppression, and exit rules.
- Do not use discounts to hide a value-realization failure.
- Do not calculate strategic LTV from an uncalibrated hypothesis tree.
- Do not subtract CAC inside LTV and then subtract it again in acquisition economics.
- Do not compare cohorts or periods with inconsistent margin, churn, or service-cost definitions.
- Do not optimize retention by trapping customers, obscuring cancellation, or damaging trust.
- Do not present Marketing Compass classifications as universal industry terminology.

Referenced files: 5

View saved version →

---
name: audit-marketing-reasoning
description: Audit marketing proposals, analyses, reports, dashboards, case studies, and frameworks for causal validity, evidence quality, reproducibility, context transfer, information freshness, ethical risk, and real-world implementability. Use when evaluating whether correlation supports causation, whether a successful case generalizes, whether metrics or models justify a decision, why a theoretically sound proposal may fail in practice, or how to rewrite a marketing claim into a falsifiable and operational form. Route measurement-system design and metric calculations primarily to design-marketing-measurement; route broad growth diagnosis primarily to diagnose-marketing-structure.
---

# Marketing Reasoning Audit

Audit consequential claims without turning the review into fault-finding. Preserve the useful core, expose uncertainty, and convert criticism into a safer decision or a small validation step.

## Operating principles

- Treat frameworks as subordinate to observed reality.
- Separate business outcomes from the proposed marketing mechanism.
- Distinguish fact, interpretation, hypothesis, unknown, and recommendation.
- Do not convert correlation, model output, or a successful case into causation by wording alone.
- Treat missing information as a reason to condition a decision, not automatically to stop it.
- Describe incentive, political, or operational risks as hypotheses unless directly evidenced.
- Protect customer trust; weak signals do not create strong permission to sell.
- Date unstable assumptions and verify current facts when the decision depends on them.

## Workflow

### 1. Define the audit object

Identify:

- the decision the material is meant to support;
- the material claims that could change that decision;
- the population, market, channel, product, period, and organization in scope;
- the cost of being wrong and the reversibility of the action.

If scope is unclear, state a provisional scope and ask only for information that could materially change the verdict.

### 2. Build a claim map

For each material claim, trace:

`evidence -> inference -> conclusion -> proposed action`

Label every element as fact, interpretation, hypothesis, unknown, or recommendation. Flag any skipped link, undefined term, proxy presented as an outcome, or meaning formula presented as a calculation formula.

Read [reasoning-audit.md](references/reasoning-audit.md) when auditing a report, case study, model, or causal claim in detail.

### 3. Audit evidence and causality

Check at minimum:

- denominator, distribution, comparison group, and time window;
- time order and a credible counterfactual;
- common causes, reverse causality, selection bias, and survivorship bias;
- measurement error, aggregation, collinearity, seasonality, and repeated testing;
- whether the evidence measures behavior, attitude, recall, revenue, or only a proxy;
- whether alternative explanations remain plausible.

Use four evidence levels:

- **A — causal:** credible experiment or strong quasi-experimental design.
- **B — triangulated:** multiple independent methods support the same direction.
- **C — observational:** association or model estimate with unresolved alternatives.
- **D — hypothesis:** expert judgment, anecdote, analogy, or untested mechanism.

Do not treat MMM as automatic causal proof. Use it primarily for hypotheses, anomaly detection, model and data audit, broad ranges, and directional risk. Pair it with experiments, Geo tests, surveys, or other evidence when the decision needs incrementality.

### 4. Audit transfer and reproducibility

Test whether the reasoning survives changes in:

- customer and buying situation;
- product category and price;
- competitive structure and distribution;
- channel, creative, and timing;
- operating capacity, incentives, and approval process.

Look for failed, neutral, and excluded cases—not only winners. A single high-CTR campaign, viral post, or successful company is evidence of possibility, not a universal rule.

### 5. Audit freshness and implementation

Read [implementation-and-freshness.md](references/implementation-and-freshness.md) when the claim depends on changing platforms, AI, tactics, internal politics, incentives, or operational capacity.

Check:

- when the source, data, and premise were last valid;
- whether the mechanism has changed since then;
- who must act, what process changes, and what capacity is required;
- local KPI conflicts, workarounds, vested interests, and approval incentives;
- whether the proposal is reversible, stageable, and observable after launch.

Do not call a plan implementable merely because its logic is sound. Equally, do not use organizational friction as an excuse to abandon a valuable hypothesis; propose staged adoption and explicit conditions.

### 6. Issue a decision and repair the claim

Choose one verdict:

- **Support:** evidence and implementation conditions justify the stated action.
- **Conditional support:** useful direction, but scope or conditions must be explicit.
- **Hold:** material unknowns make the current action premature; define the smallest test.
- **Refute:** evidence contradicts the claim or the inference is structurally invalid.

For every important weakness, provide:

1. the original risk;
2. a corrected, falsifiable wording;
3. the smallest useful validation;
4. implementation and withdrawal conditions.

Use [output-contract.md](references/output-contract.md) for the final structure.

## Routing boundaries

- Use `design-marketing-measurement` first for KPI architecture, experimental design, MMM specification, or metric calculation. Apply this skill as a second-line reasoning audit.
- Use `diagnose-marketing-structure` first when the primary question is why revenue or growth is weak and the bottleneck is unknown.
- Use `evaluate-ad-investment` first for budget, media, brand-versus-performance, or investment classification decisions.
- Use `assess-ma-crm-ltv` first for MA, CRM, retention, lifecycle, or LTV suitability.
- Use this skill first when the artifact's logic, evidence, generalizability, freshness, or implementation assumptions are themselves under review.

## Guardrails

- Do not diagnose a person's motives, competence, or psychology from a proposal.
- Do not hide behind “it depends”; name the variables on which it depends.
- Do not block reversible learning because perfect information is unavailable.
- Do not present proprietary terminology as an established industry standard.
- Do not optimize a KPI in a way that predictably damages trust or long-term choice.
- Do not erase rejected, non-buying, or churned people from stakeholder analysis.
- Do not confuse a polished audit with truth; disclose residual uncertainty.

Referenced files: 5

View saved version →

---
name: design-btob-growth
description: "Diagnose and design B2B growth across ICP and demand triggers, contact and recall, first meetings, problem resolution, comparison, buying committees, internal approval, procurement, implementation, value realization, and continuing relationships. Use when a user asks why B2B leads, meetings, opportunities, proposals, or wins are not growing; how to improve discovery or first sales meetings; why deals stall; how to follow up content downloads; how to create ROI, comparison, approval, or internal-selling materials; or how to separate automation from human intervention."
---

# Design B2B Growth

Treat B2B growth as a multi-person decision and implementation system, not a lead-volume funnel. Diagnose how the customer's organization decides, explains, funds, approves, adopts, and realizes value.

Read [references/btob-structure.md](references/btob-structure.md) for every substantive diagnosis. Read [references/meeting-and-deal-design.md](references/meeting-and-deal-design.md) when designing meetings, reviving stalled deals, or creating internal-selling assets. Read [references/output-contract.md](references/output-contract.md) before returning a complete design.

## Establish the commercial decision

Identify:

- Who on the seller side must decide what and by when
- The product, contract form, sales cycle, and implementation burden
- The target account and excluded account conditions
- The customer's need event or decision trigger
- The current stage and observed stop point
- The acceptable acquisition cost, sales capacity, and delivery capacity
- What evidence exists beyond lead or activity counts

If the target account, need trigger, or value realization is undefined, do not begin by scripting outreach or automating nurture. Ask at most three questions that can change the diagnosis, then continue provisionally if possible.

## Map the B2B decision system

Use:

```text
Demand-bearing account / ICP
  → contact and recall
  → problem resolution, comparison, and sales value
  → internal approval and consensus
  → procurement and implementation
  → value realization
  → continuing relationship
```

For each stage, separate:

- Confirmed facts
- Seller interpretation
- Buyer-side hypothesis
- Unknowns
- Observable signal and missing evidence

Select one primary bottleneck. Do not treat all stages as equal priorities.

## Define ICP and demand trigger

Specify ICP through conditions that affect value and buying ability:

- Problem and use case
- Need event or change trigger
- Operating scale and economic impact
- Current workaround and dissatisfaction
- Decision authority and budget path
- Implementation readiness and constraints
- Timing and urgency
- Exclusion conditions

Do not define ICP only by industry, company size, or job title. Distinguish an account that fits the product from an account that is ready to decide now.

## Map the buying committee

Identify roles rather than assuming one buyer:

- Problem owner
- User or operator
- Champion
- Economic buyer
- Technical, legal, security, procurement, or finance reviewer
- Executive sponsor
- Blocker or status-quo beneficiary

For each relevant role, record success criteria, perceived risk, proof required, approval power, and the language they need to explain the decision internally.

Do not mistake a friendly contact for decision authority or consensus.

## Diagnose the stop structure

Classify the primary stop point:

- No real need or wrong ICP
- Need exists but is not articulated or urgent
- Contact exists but recall or trust is weak
- Problem is understood but the proposed value is generic
- Comparison criteria are missing or unfavorable
- Economic case or budget path is absent
- Internal consensus or approval format is missing
- Procurement, security, legal, or implementation risk blocks progress
- Expected value is unclear after purchase
- The next step has no owner, date, or deliverable

Do not explain a stalled deal as buyer hesitation alone. Prefer the structural condition that prevents the next commitment.

## Redesign the first meeting

Use the first meeting to jointly increase problem resolution and create language that can travel inside the customer's organization.

Cover:

1. Why the issue matters now
2. Current process, workaround, and cost of inaction
3. Stakeholders and internal decision route
4. Desired post-adoption change
5. Evaluation and comparison criteria
6. Risks and proof required
7. The smallest credible next step

Translate features into use, operational change, business effect, and internal explanation. Limit company introduction to what establishes relevance and trust.

## Create internal-selling assets

Design only assets that remove a diagnosed blocker:

- Problem and current-state summary in buyer language
- Before/after operating picture
- Comparison table with decision criteria
- ROI or economic-case assumptions with ranges
- Risk, security, legal, procurement, and implementation answers
- Stakeholder-specific one-page summaries
- Approval memo or executive brief
- Implementation plan and responsibility map
- Proof, case, reference, or pilot design

Do not fabricate ROI or proof. Label assumptions, source, owner, and validation need.

## Handle content downloads and nurture

Treat a download as a weak signal unless additional behavior or context raises confidence. Classify the next action by temperature:

- Immediate active evaluation
- Recognized problem, timing unknown
- Future demand or internal reference
- General learning or irrelevant interest

Make content function as a quiet seed and proxy presentation: memorable framing, post-adoption change, comparison logic, expected questions, and material the reader can reuse internally.

Do not convert every download into immediate phone outreach. Use progressive permission and behavior appropriate to the signal strength.

## Separate automation and human intervention

Automate repeatable timing, routing, reminders, content delivery, and known friction. Use people for:

- Ambiguous problem definition
- Stakeholder conflict
- Economic and operational tradeoffs
- New objections or political risk
- Approval language and mutual commitments

Do not use automation to simulate relationship or pressure accounts without demand.

## Define progression and evidence

For each consequential stage, define:

```text
Entry condition
Exit condition
Buyer commitment
Seller deliverable
Desired Signal
Counter-signal
Guardrail
Review point
Action if crossed
```

Prefer buyer commitments and decision evidence over seller activities. An email sent, meeting held, proposal delivered, or content opened is not stage progression by itself.

## Return a B2B growth design

Use the compact or full structure in [references/output-contract.md](references/output-contract.md). Include:

- ICP and exclusion criteria
- Demand trigger
- Buying committee and approval route
- Primary stop structure
- Meeting or deal intervention
- Missing internal-selling asset
- Next mutual commitment
- Automation and human-intervention split
- Signal, counter-signal, guardrail, and review point
- What not to do

## Guardrails

- Do not optimize lead quantity before confirming ICP, demand, and value.
- Do not treat content downloads or opens as buying intent.
- Do not confuse seller activity with buyer progress.
- Do not force one contact to carry an organization-wide decision unsupported.
- Do not invent ROI, urgency, authority, or consensus.
- Do not respond to structural deal blockage with repeated closing alone.
- Do not promise implementation or value the delivery organization cannot support.
- Do not automate contact when human judgment or explicit permission is required.
- Do not present Marketing Compass classifications as universal industry terminology.

Referenced files: 5

View saved version →

---
name: design-marketing-communications
description: "Design marketing communications from a defined customer-state transition, meaning, recall, trust, comparison, expectation, and action path rather than from channels. Use when a user asks for a communications strategy, brand message, awareness or recall plan, PESO or touchpoint roles, TV/PR/SNS/Web/store integration, a functional-versus-emotional message decision, expectation and UX alignment, word-of-mouth or buzz design, or a route from paid contact into owned assets, search, sales, store, product, or customer success."
---

# Design Marketing Communications

Design the customer-state transition before selecting media or producing messages. Communication is one intervention within marketing; do not use it to replace product value, availability, price, implementation, or experience.

Read [references/state-and-meaning.md](references/state-and-meaning.md) for every substantive design. Read [references/touchpoint-and-transmission.md](references/touchpoint-and-transmission.md) when assigning PESO, real/digital roles, or word-of-mouth. Read [references/output-contract.md](references/output-contract.md) before returning a complete strategy.

## Establish the decision

Identify:

- Who must decide what and by when
- The customer, buying situation, and demand state
- The current customer state
- The single next state communication should create
- The behavior or downstream path that state should enable
- Product, price, distribution, experience, sales, or operational constraints
- The acceptable cost, trust risk, and review horizon

If the primary bottleneck is product value, availability, adoption, or continuation, do not recommend communication as the primary solution. Ask at most three questions that can change the state transition, then proceed provisionally when possible.

## Define one state transition

Use the smallest relevant path:

```text
Latent or unrecognized need
  → recognition or category entry
  → awareness and memory
  → recall in the buying situation
  → understanding and meaning
  → trust, preference, or consideration
  → comparison and choice
  → intended action
  → expectation and experience
  → re-recall, re-selection, or advocacy
```

Name:

- Current state
- Desired next state
- Barrier to transition
- Customer evidence that the transition occurred
- Downstream owner after communication

Do not attempt to move every state in one execution. Recall and trust may support multiple stages, but still define the primary transition.

## Diagnose the buying condition

Classify only what changes the design:

- Convenience versus specialty purchase
- Low versus high involvement
- Impulse versus planned purchase
- Individual versus organizational decision
- Immediate demand versus future demand
- Search-led, shelf-led, relationship-led, or habit-led choice

Treat these as contextual dimensions, not fixed product categories. The same offer may require different communication by use case.

## Design the meaning

Use:

```text
Product → meaning → customer context → choice
```

Define:

- The customer as subject, not the brand
- The situation in which the message should be recalled
- One meaning the customer should reconstruct
- The choice reason that meaning supports
- Proof or experience that makes it credible
- What should not be remembered or overpromised

Measure success by what was reconstructed, not what was transmitted. Reduce information before adding more. Do not use brand jargon that requires the sender to explain it.

## Balance function, story, emotion, and logic

Assign roles rather than choosing a universal winner:

- Function creates understanding and practical credibility.
- Story creates context, empathy, and motivation.
- Emotion starts or accelerates attention and action.
- Logic removes uncertainty, perceived risk, and braking forces.

Match the mix to involvement, decision risk, prior knowledge, and channel context. Do not frame function versus story or emotion versus reason as a permanent binary.

## Design recall and selection

For recall work, specify:

- Category-entry or need moment
- Memory cue
- Distinctive meaning or association
- Reach and repetition logic
- Competitive and physical availability context
- Route from recall to access or comparison

Do not equate delivered awareness with recall in a buying situation. If the offer is recalled but not chosen, work on selection reason, proof, price, risk, or access rather than more awareness alone.

## Align expectation and experience

Map the promise across advertising, PR, social proof, site, sales, store, packaging, onboarding, product, and CS.

Check:

- What expectation communication creates
- What the customer actually experiences
- Time and effort to reach value
- Where overstatement creates disappointment
- Where under-explaining prevents value recognition
- Which downstream team owns the handoff

Do not maximize conversion by creating expectations the experience cannot satisfy.

## Assign touchpoint roles

Use PESO as a role map, not a checklist:

- **Paid:** controlled contact, reach, timing, and short-term acceleration.
- **Earned:** third-party credibility, public relevance, and interpretation.
- **Shared:** peer transmission, participation, conversation, and UGC.
- **Owned:** depth, evidence, comparison, action, relationship, and reusable memory.

Assign every touchpoint one primary state transition and a handoff. Connect paid contact to owned or operational assets when continued understanding, action, or relationship matters.

Combine digital breadth and observability with analog depth, embodiment, proximity, or trust where useful. Do not default to either channel family.

## Design transmission without promising buzz

For word-of-mouth or buzz, design:

- The reason one person would give it to another
- The narrow first subject who sees themselves in it
- The existing conversation, frustration, identity, or useful task it joins
- The emotion the sharing helps process
- The social cost or ambiguity that might stop transmission
- A path for meaning to remain accurate as it travels

Treat content, timing, platform, context, and emotion as an interacting chain. Do not promise virality, reach, who will be hit, or repeatability.

## Define measurement and guardrails

Select measures that match the transition:

- Delivery: reach, exposure, frequency, completion
- State change: recognition, recall, understanding, trust, preference, consideration
- Path behavior: search, direct visit, comparison, store action, inquiry, trial
- Experience: expectation gap, activation, satisfaction, complaint, return, continuation
- Transmission: share motive, message fidelity, relevant conversation, earned response

Use:

```text
Baseline
Desired Signal
Counter-signal
Guardrail
Review window
Action if crossed
```

Do not attribute sales directly without a defensible counterfactual and control of material non-communication variables.

## Return a communications design

Use the compact or full structure in [references/output-contract.md](references/output-contract.md). Include:

- Current and desired customer state
- Barrier and intervention mechanism
- Meaning to reconstruct and proof
- Function/story/emotion/logic roles
- PESO and touchpoint handoffs
- Expectation-experience alignment
- Signal, counter-signal, guardrail, and review point
- What not to communicate or optimize

## Guardrails

- Do not select channels before defining the state transition.
- Do not use communication to cover absent product value or availability.
- Do not call exposure, impressions, or views customer-state change.
- Do not optimize immediate conversion through false scarcity, misleading framing, or expectation inflation.
- Do not equate awareness with recall, preference, or choice.
- Do not present function and story, emotion and logic, or digital and analog as fixed opposites.
- Do not promise buzz, earned coverage, UGC, or precise virality.
- Do not attribute revenue to communication without a defensible counterfactual.
- Do not present Marketing Compass classifications as universal industry terminology.

Referenced files: 5

View saved version →

---
name: design-marketing-measurement
description: "Design or audit marketing measurement systems, KPI trees, effect measurement, experiments, MMM interpretation, and continuation or withdrawal rules. Use when a user asks what to measure, how to connect a marketing objective to KGI and KPI, whether sales or ROAS is an appropriate success metric, how to choose among A/B tests, Geo tests, surveys, logs, MMM, or proxy signals, how much causal confidence a result supports, or how to define Baseline, Desired Signal, Counter-signal, Guardrail, review window, and action thresholds."
---

# Design Marketing Measurement

Design measurement for a decision, not a dashboard. Start from what the organization must decide and what return the intervention is meant to create. Do not begin with available metrics.

Read [references/measurement-principles.md](references/measurement-principles.md) for every substantive design. Read [references/method-selection.md](references/method-selection.md) when selecting evidence or auditing MMM and experiments. Read [references/output-contract.md](references/output-contract.md) before returning a complete specification.

## Establish the decision and return

Identify:

- Who will decide what, and by when
- The intervention and population in scope
- The single primary return `R`
- The expected time horizon
- The acceptable loss or downside
- The action that a result will trigger

Define `R` as the customer or business state the intervention is supposed to change: recall, understanding, preference, trust, qualified demand, purchase, activation, continuation, profit, or another explicit state.

If the marketing structure or primary bottleneck is still unknown, do not create a company-wide KPI tree. Return to structural diagnosis or produce only a provisional measurement discovery plan.

## Separate responsibility levels

Distinguish:

1. **Business KGI:** the organization-level result, such as profit, revenue, customers, or continuation.
2. **Intervention KGI:** the customer state or result the specific intervention can reasonably own.
3. **KPI:** an intermediate variable hypothesized to move the intervention KGI.
4. **Operational metric:** delivery, quality, or execution health.

Do not assign sales as the direct KGI of one communication intervention unless a defensible counterfactual and control of material product, price, distribution, competitive, seasonal, and sales effects exist.

Do not call reach, clicks, downloads, or open rates business outcomes. They can be delivery metrics, proxies, or weak signals depending on the decision.

## Build the smallest useful causal chain

Use:

```text
Intervention → customer-state change → behavior → business outcome
```

For every arrow, record:

- The causal hypothesis
- Population and period
- Known confounders
- Observable measure or proxy
- What evidence would weaken the link

Classify each expression as a meaning equation, accounting equation, probability equation, or hypothesis tree. Do not calculate a meaning equation or uncalibrated hypothesis tree.

Select a small number of decision-relevant measures. Do not reward a KPI merely because it is easy to collect.

## Choose the evidence method

Select methods by the claim required, intervention level, feasibility, lag, and risk.

- Use logs for observed behavior and delivery.
- Use surveys for recognition, recall, understanding, preference, trust, and intent.
- Use user-level A/B tests for treatment differences within a controllable interface or message.
- Use Geo or comparable market tests for macro incremental effects when spillover and region comparability can be managed.
- Use MMM for historical fit, hypotheses, anomalies, audits, and rough ranges—not causal truth or exact allocation.
- Use multiple independent signals when no single method can identify the effect.

Read `method-selection.md` for method constraints. State what the design can and cannot establish.

## Define the decision boundary

Every consequential measure must include:

```text
Baseline
Desired Signal
Counter-signal
Guardrail
Review window
Action if crossed
```

Use ranges or directional criteria when false precision would be misleading. Define the population, denominator, attribution window, cohort, data source, owner, and measurement cadence for every operationalized metric.

Separate optimization problems from threshold or defense problems. For delayed or potentially irreversible loss, use small changes, provisional floors, anomaly detection, and restoration rules rather than searching aggressively for an optimum.

## Audit causal claims

Check:

- Counterfactual quality
- Randomization or assignment integrity
- Contamination and spillover
- Seasonality and concurrent changes
- Selection, survivorship, and attrition bias
- Reverse causality
- Common causes and collinearity
- Segment averages hiding heterogeneous effects
- Multiple testing and post-hoc metric selection
- Data freshness and instrumentation changes

Assign the evidence level in `measurement-principles.md`. Use causal language only at the level the design supports.

## Handle common metric errors

- If KPI improves and KGI does not, revise the KPI or causal hypothesis.
- If ROAS is high without an incrementality design, describe efficiency or attributed return, not causal return.
- Define Incremental CPA as `additional advertising cost ÷ incremental conversions`.
- Define Incremental ROAS as `incremental revenue ÷ additional advertising cost`.
- Do not call `additional advertising cost ÷ incremental revenue` Incremental CPA.
- Do not infer latent attitude change merely because it is plausible; measure it or label it unknown.
- Do not reject an important construct merely because it is difficult to observe.

## Return a measurement specification

Use the compact or full form in [references/output-contract.md](references/output-contract.md). Include:

- The decision and primary `R`
- Business KGI, intervention KGI, KPI, and operational metrics
- Metric definitions and equation types
- Evidence method and confidence level
- What can and cannot be concluded
- Signal, counter-signal, guardrail, review window, and action
- The minimum next measurement step

## Guardrails

- Do not manufacture baselines, thresholds, lift, or power assumptions.
- Do not present a proxy as the underlying construct.
- Do not present MMM coefficients or attributed platform conversions as causal truth.
- Do not use an A/B win to claim total market incrementality.
- Do not design measurement that cannot change a decision.
- Do not increase measurement complexity beyond the value or risk of the decision.
- Do not optimize a KPI in a way that harms customer experience, trust, margin, or long-term value.
- Do not present Marketing Compass classifications as universal industry terminology.

Referenced files: 5

View saved version →

---
name: diagnose-marketing-structure
description: "Diagnose ambiguous marketing, growth, revenue, acquisition, retention, brand, channel, or go-to-market problems before recommending tactics. Use when a user asks why sales are not growing, what marketing should prioritize, whether to invest in ads/SEO/SNS/CRM/MA, how to choose among competing initiatives, or needs the business and customer-decision structure decomposed into demand, recall/acquisition, choice/value realization, availability, continuation, and relationship. Identify the maximum bottleneck, evidence gaps, what not to do, and whether a specialist measurement, advertising, B2B, CRM/LTV, communications, or reasoning audit is needed."
---

# Diagnose Marketing Structure

Treat this skill as the entry point to Marketing Compass. Diagnose the constraint before generating tactics. Do not assume the channel named by the user is the actual problem.

Read [references/core-principles.md](references/core-principles.md) for every substantive diagnosis. Read [references/diagnostic-model.md](references/diagnostic-model.md) when mapping the business, selecting equations, or comparing bottlenecks. Read [references/output-contract.md](references/output-contract.md) before returning a full diagnosis or requesting missing information.

## Establish the decision

Identify:

- Who must decide what
- The decision deadline
- Available budget or capacity
- Tolerable loss and irreversible risks
- What is outside the decision scope

If critical context is missing, ask at most three questions that could change the decision. Do not demand exhaustive discovery. When useful, continue with an explicitly provisional diagnosis.

Do not ask for a metric merely because it is conventionally available. Ask only when it distinguishes competing bottleneck hypotheses.

## Classify the demand problem

Distinguish:

1. **Base demand occurrence:** the need state or event exists independently of marketing.
2. **Demand recognition:** the customer has the problem but has not recognized or articulated it.
3. **Demand acceleration:** marketing can bring a future decision forward.
4. **Context or category creation:** marketing can create a new meaning, use case, or comparison frame, with high uncertainty.

Do not say simply that marketing can or cannot create demand. Name which of the four is in scope.

## Build the smallest useful structure

Choose the model that matches the decision. Do not fill every framework.

- Revenue accounting: trial revenue plus repeat revenue
- Measurable operating model: traffic × conversion × value × continuation
- Customer-decision model: demand → recall/acquisition → choice/value realization → continuation/relationship
- Availability model: mental availability × physical availability
- B2B model: demand-bearing account → contact/recall → problem resolution and comparison → internal approval → value realization → continuing relationship

Separate:

- Known values
- Proxy values
- Unknown values
- Meaning equations, accounting equations, probability equations, and hypothesis trees

Never calculate with a meaning equation or an uncalibrated hypothesis tree.

## Select the maximum bottleneck

Generate competing bottleneck hypotheses, then compare them using:

1. Expected effect on the total outcome
2. Proximity to zero or a hard constraint
3. Evidence strength
4. Ability to influence the variable
5. Time to learn
6. Reversibility and downside
7. Implementation and organizational feasibility

Select one primary bottleneck and, when needed, one secondary hypothesis. If evidence cannot distinguish them, recommend the smallest test that can.

Do not select a variable merely because it is easiest to measure. Do not list many equal priorities.

## Test the customer decision

Check only the dimensions relevant to the suspected bottleneck:

- Trigger or need state
- Recall and consideration set
- Choice criteria and preference
- Price, trust, proof, and perceived risk
- Physical and digital availability
- Purchase or adoption friction
- Expectation versus experienced value
- Re-purchase, re-selection, or continuation
- For B2B: stakeholders, approval, comparison, budget, urgency, and internal explanation

Treat communications as one possible intervention, not marketing as a whole. Do not use promotion to compensate for absent value, availability, or value realization.

## Determine the evidence status

Separate facts, interpretations, hypotheses, and unknowns. Assign the evidence level in `core-principles.md`.

Check for:

- Confounding and reverse causality
- Selection and survivorship bias
- Missing attitudinal change because only behavior is logged
- Averages hiding segments
- A copied success case with different conditions
- Stale platform, market, or organizational assumptions

When only observational evidence exists, describe consistency or a plausible hypothesis, not causal proof.

## Decide and route

Return one of:

- Execute
- Small reversible test
- Hold pending evidence
- Reduce or contain
- Stop or withdraw

Route to a specialist domain only after stating why:

- Measurement, KPI, MMM, experiments, Signals/Guardrails → marketing measurement
- Current-value, non-current-value, structural advertising → advertising investment
- ICP, deals, meetings, internal approval, downloaded materials → B2B growth
- MA readiness, strategic/operational CRM, CS, LTV → MA/CRM/LTV
- Recall, meaning, PESO, expectations, touchpoint roles, buzz → marketing communications
- Causal claims, data quality, stale assumptions, organizational self-sabotage → marketing reasoning audit

If no specialist skill is available, still state the needed specialist analysis rather than fabricating it.

## Return a decision, not a framework dump

Use the compact output in [references/output-contract.md](references/output-contract.md). Include:

- The core diagnosis first
- Primary bottleneck and competing hypothesis
- Evidence and uncertainty
- Next action and what not to do
- Signal, counter-signal, guardrail, and review point when a real decision is being made

Adapt depth to the request. Do not force every field into a simple answer.

## Guardrails

- Do not attribute revenue change to one tactic without a defensible counterfactual.
- Do not invent numbers to complete a model.
- Do not call a proxy the underlying construct.
- Do not recommend large execution before locating the bottleneck.
- Do not treat a weak signal such as a content download as purchase intent.
- Do not use MA or CRM to replace demand, ICP, value, or competitive advantage.
- Do not optimize an irreversible threshold problem as if it were a smooth response curve.
- Do not confuse a logically elegant plan with one that the operating organization can execute.
- Do not present Marketing Compass classifications as universal industry terminology.

Referenced files: 5

View saved version →

---
name: evaluate-ad-investment
description: "Evaluate whether an advertising plan, channel, budget, brand campaign, performance campaign, TV campaign, or media reduction qualifies as an investment and under what conditions. Use when a user asks whether advertising is worth it, how to compare brand and performance advertising, whether to cut or increase media, what return and time horizon to use, how to classify current-value, non-current-value, or organizational advertising, or how to define incremental return, residual value, safe floors, signals, guardrails, and withdrawal rules."
---

# Evaluate Advertising Investment

Judge the investment role before judging media efficiency. Treat advertising as an intervention in the timing or probability of purchase, recall, consideration, trust, or selection. Do not assume that all advertising has the same return period or can be compared by one metric.

Read [references/investment-model.md](references/investment-model.md) for every substantive evaluation. Read [references/decision-branches.md](references/decision-branches.md) when distinguishing optimization from threshold defense or selecting tests. Read [references/output-contract.md](references/output-contract.md) before returning a complete recommendation.

## Establish the decision

Identify:

- Who must decide what and by when
- The product, market, customer, and buying situation
- The budget change under consideration
- The probability or timing the advertising is intended to change
- The evaluation period and expected lag
- The acceptable loss, recovery time, and irreversible risk
- Product, price, distribution, sales, experience, or demand constraints outside advertising

If the underlying sales problem or primary bottleneck is unknown, do not approve a large advertising change. Return to structural diagnosis or limit the answer to a provisional test.

Ask no more than three questions that can change the classification or decision. Continue with labeled assumptions when appropriate.

## Name the intended probability

Choose the primary intervention target:

- Bring forward an already-likely purchase
- Enter or strengthen recall at a future category-entry moment
- Enter the consideration set
- Increase trust or reduce perceived risk
- Change meaning, comparison frame, or category context
- Maintain a baseline against forgetting or competitive pressure

Do not use “awareness” as a complete objective. State what awareness is expected to enable.

## Classify the advertising

Assign one primary class from `investment-model.md`:

1. **Current-value advertising:** buys time by accelerating existing demand through a known recovery path.
2. **Non-current-value advertising:** buys a future option or probability such as recall, trust, or consideration entry.
3. **Organizational advertising:** primarily buys internal reassurance, legitimacy, continuity, relationships, or institutional convenience while the market probability remains undefined.

Use `defensive / maintenance cost` only as an operating tag under non-current-value advertising; do not create a fourth class.

If one plan has materially different roles, split it into components rather than forcing a blended label.

## Test investment qualification

For current-value advertising, require:

- Existing or identifiable demand
- A defined conversion and profit path
- Incremental or marginal return logic
- Capacity, stock, sales, and value realization
- A limit where additional spend no longer qualifies
- Expected behavior when spend stops

For non-current-value advertising, require:

- A defined future customer-state change
- A plausible mechanism and relevant buying horizon
- Leading signals and counter-signals
- A theory of what persists after exposure
- A review period matched to lag
- A withdrawal or redesign rule

For organizational advertising, require transparent accounting as an organizational cost. Do not disguise it as market investment or assign invented ROI.

## Separate optimization from defense

Treat a response as an optimization problem only when changes are sufficiently observable, continuous, timely, and reversible.

Treat it as a threshold or defense problem when:

- Effects are delayed or small relative to noise
- Loss may appear only after memory, distribution, or competitive position erodes
- Recovery is slow or uncertain
- Large cuts may cross a hidden floor

For defense problems, define a provisional safe floor, make small changes, monitor anomalies, and restore when the guardrail is crossed. Do not cut until sales reveal the damage.

## Define return and evidence

Match return to the classification:

- Current value: incremental conversions, contribution profit, Incremental CPA, Incremental ROAS, marginal return, payback.
- Non-current value: recall, consideration entry, qualified direct demand, trust, future conversion quality, residual value, and independent business signals.
- Defensive maintenance: stability bands, share and distribution context, competitive pressure, brand search or recall anomalies, and recovery risk.
- Organizational: explicit organizational benefit and cost, separate from market ROI.

Define:

```text
Incremental effect = observed result − counterfactual result
Incremental CPA = additional advertising cost ÷ incremental conversions
Incremental ROAS = incremental revenue ÷ additional advertising cost
```

Do not treat attributed platform return, correlation, or MMM coefficients as causal truth. State the evidence level and what can and cannot be concluded.

## Design the decision boundary

Specify:

```text
Baseline
Desired Signal
Counter-signal
Guardrail
Review window
Action if crossed
```

For each threshold, state whether it is evidence-based, a provisional heuristic, or unknown. Use ranges or directional criteria when precision is unjustified.

Choose one decision state:

- Increase
- Continue
- Small reversible test
- Hold
- Reduce
- Stop or restore

## Return an investment judgment

Use the compact or full structure in [references/output-contract.md](references/output-contract.md). Include:

- Intended probability or timing change
- Advertising classification and operating tag
- Investment qualification and missing conditions
- Optimization versus threshold-defense status
- Short- and long-horizon return measures
- Residual value or stopping behavior
- Evidence level and uncertainty
- Signal, counter-signal, guardrail, review window, and action
- What not to do

## Guardrails

- Do not recommend advertising to compensate for absent demand, product value, availability, activation, or continuation.
- Do not judge all advertising by immediate ROAS.
- Do not call weak visibility metrics residual value without a mechanism.
- Do not compare brand and performance advertising as if they buy the same outcome and time horizon.
- Do not optimize a potentially irreversible brand or distribution threshold through large cuts.
- Do not present organizational advertising as a market investment.
- Do not approve a budget without a return definition, review period, and withdrawal rule.
- Do not claim causality without a defensible counterfactual.
- Do not present Marketing Compass classifications as universal industry terminology.

Referenced files: 5