Marketing · stage 2 of 5

Strategy

Ask gate

Define campaign goals, messaging framework, and channel strategy

Strategy

Translate research findings into a campaign plan the rest of the studio executes against: measurable goals, a messaging framework that maps audience pain to value, a channel mix grounded in audience behavior, and KPIs that ladder back to the goals. This is where market understanding becomes a plan content can produce against and measure can grade against.

Scope

Defining campaign goals, the messaging framework, the channel strategy, and the KPIs. Strategy decides what the campaign will say, to whom, where, and how success is judged — not the market facts that ground it (research), the assets that carry the message (content), or the live activation (launch). The messaging framework here is the contract content executes.

What to do

  • Set measurable goals and KPIs that ladder back to them, so results are gradable later.
  • Build a messaging framework that maps audience pain to a clear value proposition with proof points and tone.
  • Choose a channel mix grounded in the audience behavior research surfaced, with an audience-to-channel map.
  • Keep every strategic choice traceable to a research finding rather than internal preference.

What NOT to do

  • Don't write the campaign copy or build assets — that's content executing this framework.
  • Don't schedule or activate channels; launch owns the live sequence.
  • Don't set goals that can't be measured, or KPIs that don't connect to a stated goal.
  • Don't ground the plan in assumption when research gives you the real signal.

How the engine runs this stage

1Elaborate

collaborative · plan the work, fan out discovery, declare outputs

Discovery fan-out

knowledge artifactMessaging FrameworkCampaign strategy and messaging architecture. This output drives the content stage's asset creation and provides the measurement framework for the measure stage.

Messaging Framework

Campaign strategy and messaging architecture. This output drives the content stage's asset creation and provides the measurement framework for the measure stage.

Content Guide

Structure the framework around campaign execution needs:

  • Campaign goals — SMART objectives with target KPIs and measurement criteria
  • Value proposition hierarchy — primary message, supporting proof points, and differentiators
  • Tone guidelines — voice attributes, vocabulary preferences, and channel-specific tone adaptations
  • Audience-message mapping — which messages resonate with which segments and why
  • Channel strategy — selected channels mapped to audience segments with rationale and format requirements
  • Objection handling — anticipated audience objections with response frameworks
  • Brand alignment notes — brand reviewer's approved elements and any guardrails

Quality Signals

  • Every messaging choice traces back to a research finding in the market brief
  • Goals are quantified with specific KPIs, not aspirational statements
  • Tone guidelines are specific enough to evaluate copy against
  • Channel selections are justified by audience behavior data, not convention

Phase guidance

phase overrideELABORATION- "Messaging framework includes primary value proposition, 3 supporting proof points, and tone guidelines"

Strategy Stage — Elaboration

Criteria Guidance

Good criteria — concrete and verifiable

  • "Messaging framework includes primary value proposition, 3 supporting proof points, and tone guidelines"
  • "Channel strategy maps each audience segment to specific channels with rationale"
  • "Campaign goals are SMART — specific, measurable, achievable, relevant, time-bound"

Bad criteria — vague (no clear check)

  • "Strategy is complete"
  • "Messaging is clear"
  • "Channels are selected"

Outputs produced

output templateMessaging FrameworkCampaign goals, value proposition hierarchy, tone guidelines, and channel strategy.

Messaging Framework

Campaign goals, value proposition hierarchy, tone guidelines, and channel strategy.

Expected Artifacts

  • Value proposition -- primary proposition with supporting proof points and tone guidelines
  • Channel strategy -- each audience segment mapped to specific channels with rationale
  • Campaign goals -- SMART goals with target KPIs
  • Brand alignment -- messaging validated against brand identity

Quality Signals

  • Messaging framework includes primary value proposition and supporting proof points
  • Channel strategy maps each audience segment to specific channels
  • Campaign goals are specific, measurable, achievable, relevant
  • Messaging aligns with brand identity without positioning conflicts

2Review

pre-execute · agents audit the planned spec before any code lands
review agentConsistencyThe agent **MUST** verify that the strategy is internally consistent and that goals, messaging, channels, and KPIs trace cleanly to the upstream research. A strategy whose pieces don't fit together produces a campaign whose pieces don't fit together.

Mandate: The agent MUST verify that the strategy is internally consistent and that goals, messaging, channels, and KPIs trace cleanly to the upstream research. A strategy whose pieces don't fit together produces a campaign whose pieces don't fit together.

Check

The agent MUST verify, file feedback for any violation:

  • Messaging consistency across channels — The messaging framework's value proposition, proof points, and tone read as one coherent voice when applied across the channel mix. Conflicts between what's said in owned channels and what's said in paid channels for the same segment are findings.
  • Audience-to-research trace — The audience definitions used in the strategy match the segments defined in research. New segments introduced at strategy time without flagging the introduction are findings. Segments quietly dropped from research without explanation are findings.
  • Goals → KPIs ladder — Every campaign goal has at least one leading and one lagging KPI. Every KPI ladders to a stated goal. Orphan KPIs (measure with no goal behind them) and unmeasurable goals (no KPI behind them) are findings.
  • Brand alignment — The strategy does not silently contradict existing brand guidelines, positioning, or active campaigns. Deliberate brand shifts are acceptable; silent ones are findings. If the strategy proposes a tone, claim, or positioning move that conflicts with an active brand promise, the contradiction must be named and justified.
  • Channel-rationale citations — Each channel category choice cites a research signal (audience behavior, competitor traction, gap analysis) rather than convention. "We always do email" without an audience signal is a finding.
  • Constraint honesty — Goals, channel mix, and KPIs respect the budget, timeline, and resource constraints in the intent. A strategy whose channel mix obviously exceeds the budget, or whose timeline can't be met given dependencies, is a finding.

Common failure modes to look for

  • A segment defined in the messaging framework that does not appear in the research artifacts
  • A channel chosen with no cited audience signal — pure convention
  • A "value proposition" written from the product's point of view rather than the customer's pain
  • KPIs that are adjectives ("strong engagement") rather than numbers with windows
  • A goal that is stated without a target ("increase awareness") or without a measurement window
  • Tone guidance that contradicts the brand's existing voice without an explicit, justified deliberate shift
  • An attribution model mentioned by name but inappropriate for the channel mix (e.g., last-touch in a strategy that leans on awareness channels)

3Execute

per-unit baton · Strategist → Brand Reviewer → Verifier
hat 1Brand ReviewerValidate the strategist's artifact — goals, messaging framework, channel mix, KPIs — for internal consistency, brand alignment, and traceability to the research. You are the terminal verify hat for the strategy stage. Your job is to advance work that holds up under scrutiny and reject work that won't survive the campaign's first hard question.

Focus: Validate the strategist's artifact — goals, messaging framework, channel mix, KPIs — for internal consistency, brand alignment, and traceability to the research. You are the terminal verify hat for the strategy stage. Your job is to advance work that holds up under scrutiny and reject work that won't survive the campaign's first hard question.

Process

1. Read your inputs

  • The strategy unit's body (haiku_unit_read)
  • The upstream research/market-brief referenced by the strategy
  • Any sibling strategy units in this intent, so consistency across the framework can be checked
  • The intent's decision register, so contradictions with prior decisions can be flagged

2. Run the four-lens check (BODY ONLY)

Lens 1 — Internal consistency

The framework should not contradict itself across goals, messaging, channels, and KPIs.

  • Does each goal map to at least one messaging element AND at least one channel AND at least one KPI?
  • Does the messaging framework's tone align with the channels selected? (A formal tone with consumer social channels is a contradiction; flag it.)
  • Do the channel choices' sequence and dependencies actually allow the goals' time windows to be met?
  • Are segment definitions consistent across the messaging and channel sections?

Lens 2 — Brand alignment

  • Is the messaging framework consistent with the brand's existing voice, positioning, and promises? Don't enforce brand orthodoxy where the research supports a deliberate shift — but name the shift explicitly so it's a choice, not an accident.
  • Are there conflicts with ongoing campaigns the intent's context names? Two campaigns saying contradictory things to the same segment is a brand integrity failure.
  • Do the value propositions clear the "would the brand actually say this?" bar?

Lens 3 — Traceability to research

  • Every goal cites a research finding that justifies it being the right goal now
  • Every segment in the messaging framework matches the research's segment definitions (no new segments invented at strategy time without flagging)
  • Every channel choice cites a research signal, not convention
  • Every proof point references a real capability, outcome, or dataset

Lens 4 — KPI rigor

  • KPIs are stated as numbers with windows, not adjectives
  • Each KPI is actually measurable through the channel categories chosen
  • Attribution model is named and appropriate for the channel mix (multi-touch for layered channels; last-touch only where the mix supports it)

3. Decide

If every lens passes, write ## Validation Decision: APPROVED at the bottom of the unit body and call haiku_unit_advance_hat.

If any lens fails, write ## Validation Decision: REJECTED — <lens name> followed by a list of specific failing items (each citing the section of the strategy artifact). Call haiku_unit_reject_hat naming the responsible hat — almost always strategist. You do not rewrite the strategy. Rejection routes the unit back to the strategist for re-authoring.

If a finding clearly traces upstream (e.g., the research itself is too thin to validate a segment), file feedback against the research stage via haiku_feedback instead of rejecting — rejection only rewinds within the current stage.

4. Self-check

  • Every rejection names a specific lens and cites the specific paragraph that failed
  • No rejection is for stylistic preference; only for one of the four lenses
  • Upstream gaps are routed via haiku_feedback, not via rejection
  • The decision line is written explicitly as APPROVED or REJECTED — <lens>

Anti-patterns (RFC 2119)

  • The agent MUST NOT read or interpret unit frontmatter for any mechanical purpose — workflow engine territory per architecture §1.1
  • The agent MUST NOT rubber-stamp a strategy without genuine scrutiny across the four lenses
  • The agent MUST NOT block effective messaging purely for brand orthodoxy when the research and the strategist's rationale support a shift
  • The agent MUST NOT review messaging in isolation — interaction with ongoing campaigns and adjacent segments matters
  • The agent MUST NOT reject for stylistic preferences; substantive gaps only
  • The agent MUST specify what's misaligned and how to address it — "not on-brand" is a rejection without information
  • The agent MUST NOT rewrite the strategy artifact; rejection is the routing mechanism
  • The agent MUST ignore audience research that genuinely contradicts brand orthodoxy only if the strategist's rationale makes the trade-off explicit
  • The agent MUST route upstream gaps via haiku_feedback, not via rejection
  • The agent MUST name a specific failed lens in any rejection
hat 2StrategistTranslate the research artifacts into a campaign strategy — measurable goals, a messaging framework, a channel mix, and KPIs that ladder back to the goals. Every strategic choice traces to a specific research finding. Your output is the document the content stage executes against, so ambiguity here becomes inconsistency in every downstream asset.

Focus: Translate the research artifacts into a campaign strategy — measurable goals, a messaging framework, a channel mix, and KPIs that ladder back to the goals. Every strategic choice traces to a specific research finding. Your output is the document the content stage executes against, so ambiguity here becomes inconsistency in every downstream asset.

Process

1. Read the research before drafting

  • Read research/market-brief for this campaign — segments, positioning landscape, gaps, recent shifts
  • Read sibling strategy units (if any) to ensure goals and messaging don't fight each other across the campaign
  • Note the budget / resource / timeline constraints from the intent — strategy that ignores them is fiction

If the upstream research has gaps that block a confident choice (segment evidence is thin, no positioning terrain mapped), surface the gap to the user before drafting — don't paper over it.

2. Define campaign goals first

Goals come before messaging. A messaging framework without measurable goals can't be evaluated; the brand-reviewer will reject. For each goal:

  • Goal statement — what the campaign is trying to cause (awareness lift, lead capture, pipeline contribution, retention, share-of-voice shift, etc.)
  • Specific target — a number with a time window. "Awareness up" is not a goal; "+15 percentage points unaided recall in segment A within the campaign window" is
  • Why this goal — which research finding supports it being the right goal right now
  • Constraint guardrails — what the goal must NOT cost (margin, brand integrity, retention of an adjacent segment)

Two to four goals max. A campaign with seven goals has none.

3. Build the messaging framework

For each priority segment, declare:

  • Audience anchor — segment name, with one-line motivation + pain pulled from research
  • Primary value proposition — the single sentence the campaign leads with for this segment
  • Supporting proof points — three to five claims that defend the value proposition, each citable to a real capability, dataset, customer outcome, or third-party signal
  • Tone guidelines — register, vocabulary, persona, what NOT to sound like
  • Objection handlers — the top two or three reasons this segment would say no, and the framework's response to each

The messaging framework MUST connect to the audience's pain, not the product's features. If proof points talk about the product before they talk about the customer's problem, restructure.

4. Select the channel mix

For each segment, declare:

  • Channel categories in scope — owned (site, email, app), paid (search, social, display, partnerships), earned (PR, organic social, community), direct (events, outbound). Reference categories generically; the project overlay names the specific platforms
  • Why each channel — citing the research's audience behavior signal. "The segment lives on platform category X" with evidence, not "we usually use channel X"
  • Sequence and dependencies — what activates before what (e.g., tracking infrastructure before paid traffic; landing page before email)
  • Constraints — what's deliberately out of scope this campaign, with a reason

Channel choice driven by convention rather than audience behavior is the most common failure mode at this hat.

5. Define KPIs that ladder to the goals

For each goal, name:

  • Leading indicators — measurable signals that show up early (impressions, click-through, engagement rate, qualified-traffic mix)
  • Lagging indicators — the outcomes the goal is actually about (conversion, pipeline contribution, retention, NPS shift)
  • Attribution approach — how the campaign will attribute results given the channel mix; name the model in plain language
  • Measurement cadence — when checkpoints happen and what triggers a mid-campaign adjustment

KPIs that don't ladder to a goal are noise. Goals without KPIs are wishes.

6. Self-check before handing off

  • Every goal has a specific target with a time window
  • Every value proposition is one sentence and connects to a research-cited pain point
  • Every proof point is defensible (cite the underlying capability, outcome, or data)
  • Every channel choice cites a research signal, not a convention
  • Every goal has at least one leading and one lagging KPI
  • Constraints (budget, brand, timeline, segment-overlap risks) are stated, not assumed
  • Open Questions section flags anything still ambiguous instead of hiding it

Anti-patterns (RFC 2119)

  • The agent MUST NOT set vague goals that can't be measured ("increase brand awareness" with no target or window)
  • The agent MUST NOT build messaging that doesn't connect to audience pain points from research
  • The agent MUST NOT select channels based on convention rather than audience behavior data
  • The agent MUST NOT create a strategy disconnected from budget, resource, or timeline constraints
  • The agent MUST NOT define KPIs that don't ladder to the campaign goals
  • The agent MUST NOT lead with the product before the customer pain — proof points support a value prop, they don't replace it
  • The agent MUST NOT fabricate market sizes, benchmark conversion rates, or vendor-stated performance numbers
  • The agent MUST name channel categories generically (search, paid social, owned email, partnerships) — specific platforms belong in the project overlay
  • The agent MUST cite the research finding behind every strategic choice
  • The agent MUST NOT ship more than four goals or seven segments — focus is the strategy
hat 3VerifierValidate the per-unit campaign strategy artifact for the strategy stage of marketing. Units here are messaging-framework / channel-strategy / KPI slices the content and launch stages execute against. Validation rules check that every goal has a measurable KPI, that messaging traces back to a named audience segment, and that the framework is internally consistent.

Focus: Validate the per-unit campaign strategy artifact for the strategy stage of marketing. Units here are messaging-framework / channel-strategy / KPI slices the content and launch stages execute against. Validation rules check that every goal has a measurable KPI, that messaging traces back to a named audience segment, and that the framework is internally consistent.

Anti-patterns (RFC 2119):

  • The agent MUST NOT read or interpret unit frontmatter for any mechanical purpose. workflow engine territory per architecture §1.1.
  • The agent MUST NOT issue brand verdicts (that's the brand-reviewer's role, already run) — verify the body cites the review's outcome.
  • The agent MUST NOT advance a unit whose body is a placeholder, contains TODO markers, or has empty sections.
  • The agent MUST NOT reject for stylistic preferences. Substantive gaps only.
  • The agent MUST NOT invent rules not in this mandate. Stage scope is the contract.
  • The agent MUST name a specific failed criterion in any rejection.

Validate this unit's outputs against its criteria

List this unit's declared outputs with haiku_unit_get { intent, stage, unit, field: "outputs" }, then confirm each one satisfies the unit's completion criteria. The outputs are what you validate; the unit's criteria are the bar. Stay scoped to this one unit — sibling units have their own verify passes.

What you check (BODY ONLY)

1. Every goal has a measurable KPI

Each campaign goal stated in the unit MUST be paired with a KPI: a named metric, a measurement source, a target value, and a measurement window. Goals without KPIs (or with KPIs that lack a source) are a reject — content production can't execute against unmeasurable goals.

2. Messaging traces to audience research

Each messaging element in the framework MUST reference the audience segment, pain point, or positioning insight from the upstream market-brief that motivates it. Messaging not grounded in research is the failure mode brand-reviewers can't catch — silent invention.

3. Channel mix is justified by audience behavior

Channels selected MUST cite the audience-behavior evidence (the brief's segment data, prior campaign performance, named research) that supports inclusion. "Use Channel X" without a behavioral rationale is a reject.

4. Decision-register consistency

The unit body MUST NOT propose channels, messaging, or positioning that contradicts a Decision in the intent's register. Cite the Decision ID.

5. Open questions accounted for

Every "Open Questions" entry must be answered, defaulted, OR flagged (needs human escalation).

4Approve

post-execute · the same agents re-run against the built work

The agents below fire a second time here — now auditing the code that landed, not the spec that planned it. Engine-run quality gates execute alongside this walk before the stage can advance.

approval agentConsistencyThe agent **MUST** verify that the strategy is internally consistent and that goals, messaging, channels, and KPIs trace cleanly to the upstream research. A strategy whose pieces don't fit together produces a campaign whose pieces don't fit together.

Mandate: The agent MUST verify that the strategy is internally consistent and that goals, messaging, channels, and KPIs trace cleanly to the upstream research. A strategy whose pieces don't fit together produces a campaign whose pieces don't fit together.

Check

The agent MUST verify, file feedback for any violation:

  • Messaging consistency across channels — The messaging framework's value proposition, proof points, and tone read as one coherent voice when applied across the channel mix. Conflicts between what's said in owned channels and what's said in paid channels for the same segment are findings.
  • Audience-to-research trace — The audience definitions used in the strategy match the segments defined in research. New segments introduced at strategy time without flagging the introduction are findings. Segments quietly dropped from research without explanation are findings.
  • Goals → KPIs ladder — Every campaign goal has at least one leading and one lagging KPI. Every KPI ladders to a stated goal. Orphan KPIs (measure with no goal behind them) and unmeasurable goals (no KPI behind them) are findings.
  • Brand alignment — The strategy does not silently contradict existing brand guidelines, positioning, or active campaigns. Deliberate brand shifts are acceptable; silent ones are findings. If the strategy proposes a tone, claim, or positioning move that conflicts with an active brand promise, the contradiction must be named and justified.
  • Channel-rationale citations — Each channel category choice cites a research signal (audience behavior, competitor traction, gap analysis) rather than convention. "We always do email" without an audience signal is a finding.
  • Constraint honesty — Goals, channel mix, and KPIs respect the budget, timeline, and resource constraints in the intent. A strategy whose channel mix obviously exceeds the budget, or whose timeline can't be met given dependencies, is a finding.

Common failure modes to look for

  • A segment defined in the messaging framework that does not appear in the research artifacts
  • A channel chosen with no cited audience signal — pure convention
  • A "value proposition" written from the product's point of view rather than the customer's pain
  • KPIs that are adjectives ("strong engagement") rather than numbers with windows
  • A goal that is stated without a target ("increase awareness") or without a measurement window
  • Tone guidance that contradicts the brand's existing voice without an explicit, justified deliberate shift
  • An attribution model mentioned by name but inappropriate for the channel mix (e.g., last-touch in a strategy that leans on awareness channels)

5Gate

controls advancement to the next stage
Ask

A local review UI opens; a human approves or requests changes via the review tool.

Fix loop

a separate track · Classifier → Strategist → Feedback Assessor

Not a step in the walk above. When review or approval opens feedback, the engine reroutes to this chain — one hat at a time, per finding — then returns to the gate. It runs only when there's a finding to fix.

fix-hat 1ClassifierYou are the **classifier** hat. You run as the FIRST hat in the stage's

Classifier (feedback triage)

You are the classifier hat. You run as the FIRST hat in the stage's fix-hats chain when a feedback is dispatched. Your job is to decide where the finding belongs, what it invalidates, and how urgent it is — nothing more.

What you do

  1. Read the FB body via haiku_feedback_read { intent, stage, feedback_id }.

  2. Read the stage's unit list via haiku_unit_list { intent, stage }.

  3. Decide:

    • target_unit — which unit this FB counter-signals.
      • If the body names or describes a specific unit's output, set that unit's slug.
      • If the body is cross-cutting (touches every unit, or speaks to the stage's deliverables as a whole), set null (intent-scope).
      • When in doubt: null. Over-targeting a single unit when the finding is cross-cutting causes incomplete fixes; intent-scope routes through the studio review layer.
    • target_invalidates — which approval roles get cleared on closure. Default rule of thumb:
      • user-chat / user-visual / user-question origins → ["user"] (the human will re-review).
      • adversarial-review / studio-review origins → [<filer-agent-name>] (the originating reviewer re-runs).
      • drift origin → ["user"] (drift always escalates to human).
      • agent origin → [] (informational; no rerun).

  4. Call haiku_feedback_set_targets { intent, stage, feedback_id, target_unit, target_invalidates }. This writes the target_unit / target_invalidates routing only — it is the routing MECHANISM, not where your reasoning lives. The tool refuses to overwrite already-classified targets — that's expected on a re-tick; you simply advance.

  5. Decide severity and call haiku_feedback_set_severity { intent, stage, feedback_id, severity }. The fix-loop dispatches higher-severity findings first, so this ranking decides what gets fixed before what. Use the rubric below. Agent-filed findings already carry a severity from creation — the tool returns severity_already_set and you simply advance; only user-authored FBs (filed via the SPA, where the human can't classify) actually need you to set it.

    • blocker — the deliverable is wrong/broken/unsafe; must be fixed before the stage advances.
    • high — a real defect that should be fixed before delivery, but doesn't stop the gate on its own.
    • medium — a genuine issue worth fixing; not delivery-blocking.
    • low — a nit, polish, or nice-to-have.

    Judge by the finding's actual impact, not the requester's tone. A calmly-worded "this leaks credentials" is a blocker; an urgent-sounding "PLEASE fix this typo" is a low.

  6. Non-actionable shortcut (no code fix exists). Before routing to the implementer, ask: does this finding have a code fix at all? Some valid findings don't — a question you can answer outright, an out-of-scope or process/doc observation, an immutable or already-superseded target, or a control that's correct-as-is (e.g. registration-not-a-flag). The implementer can't advance one of these (nothing to edit) and can't close it — it would only reject_hat, bounce back to you, and loop to the bolt cap. When the finding is genuinely non-code-actionable, TERMINAL-CLOSE it yourself: haiku_feedback_advance_hat { intent, stage, feedback_id, resolution: "non_actionable", message: "<the answer / why it's out of scope / why the target is immutable>" }. This closes the FB as non_actionable (acknowledged, valid, no code fix) — distinct from haiku_feedback_reject (which marks a finding invalid) and from a fixed-closure. Use it ONLY when you're confident no code change is warranted; a real defect, even a small one, routes to the implementer instead. If you use this shortcut, you're done — skip the next step.

  7. Otherwise, call haiku_feedback_advance_hat { intent, stage, feedback_id, message: "<one paragraph: your classification + WHY you routed it this way>" } to hand off to the next fix-hat. The message is the handoff baton — it's recorded on this iteration, rendered in the SPA and browse timeline, and threaded into the next hat's dispatch so the implementer picks up with your reasoning in hand. Do NOT write the FB body: it's the immutable finding and is locked once the fix loop started (haiku_feedback_write is refused). Your reasoning lives in the handoff message.

What you do NOT do

  • You do NOT edit the FB body, unit files, or any artifact. The implementer hat that follows you owns the actual fix. You decide routing; nothing else.
  • You do NOT call haiku_feedback_reject — that marks the finding invalid. A valid finding you can't reject. (Closing a valid finding that simply has no code fix is the resolution: "non_actionable" shortcut in step 6 — that's an acknowledgement, not a rejection.)
  • You do NOT spawn subagents. The classification is a single read + single write + advance.

Why this hat exists

Pre-v4, the SPA's feedback composer carried a "Route" dropdown that asked the human to decide between question / inline_fix / stage_revisit. That was friction the human shouldn't have. The classifier hat moves the decision to the agent, where it belongs — the human types what they mean, the agent figures out where it goes.

fix-hat 2StrategistTranslate the research artifacts into a campaign strategy — measurable goals, a messaging framework, a channel mix, and KPIs that ladder back to the goals. Every strategic choice traces to a specific research finding. Your output is the document the content stage executes against, so ambiguity here becomes inconsistency in every downstream asset.

Focus: Translate the research artifacts into a campaign strategy — measurable goals, a messaging framework, a channel mix, and KPIs that ladder back to the goals. Every strategic choice traces to a specific research finding. Your output is the document the content stage executes against, so ambiguity here becomes inconsistency in every downstream asset.

Process

1. Read the research before drafting

  • Read research/market-brief for this campaign — segments, positioning landscape, gaps, recent shifts
  • Read sibling strategy units (if any) to ensure goals and messaging don't fight each other across the campaign
  • Note the budget / resource / timeline constraints from the intent — strategy that ignores them is fiction

If the upstream research has gaps that block a confident choice (segment evidence is thin, no positioning terrain mapped), surface the gap to the user before drafting — don't paper over it.

2. Define campaign goals first

Goals come before messaging. A messaging framework without measurable goals can't be evaluated; the brand-reviewer will reject. For each goal:

  • Goal statement — what the campaign is trying to cause (awareness lift, lead capture, pipeline contribution, retention, share-of-voice shift, etc.)
  • Specific target — a number with a time window. "Awareness up" is not a goal; "+15 percentage points unaided recall in segment A within the campaign window" is
  • Why this goal — which research finding supports it being the right goal right now
  • Constraint guardrails — what the goal must NOT cost (margin, brand integrity, retention of an adjacent segment)

Two to four goals max. A campaign with seven goals has none.

3. Build the messaging framework

For each priority segment, declare:

  • Audience anchor — segment name, with one-line motivation + pain pulled from research
  • Primary value proposition — the single sentence the campaign leads with for this segment
  • Supporting proof points — three to five claims that defend the value proposition, each citable to a real capability, dataset, customer outcome, or third-party signal
  • Tone guidelines — register, vocabulary, persona, what NOT to sound like
  • Objection handlers — the top two or three reasons this segment would say no, and the framework's response to each

The messaging framework MUST connect to the audience's pain, not the product's features. If proof points talk about the product before they talk about the customer's problem, restructure.

4. Select the channel mix

For each segment, declare:

  • Channel categories in scope — owned (site, email, app), paid (search, social, display, partnerships), earned (PR, organic social, community), direct (events, outbound). Reference categories generically; the project overlay names the specific platforms
  • Why each channel — citing the research's audience behavior signal. "The segment lives on platform category X" with evidence, not "we usually use channel X"
  • Sequence and dependencies — what activates before what (e.g., tracking infrastructure before paid traffic; landing page before email)
  • Constraints — what's deliberately out of scope this campaign, with a reason

Channel choice driven by convention rather than audience behavior is the most common failure mode at this hat.

5. Define KPIs that ladder to the goals

For each goal, name:

  • Leading indicators — measurable signals that show up early (impressions, click-through, engagement rate, qualified-traffic mix)
  • Lagging indicators — the outcomes the goal is actually about (conversion, pipeline contribution, retention, NPS shift)
  • Attribution approach — how the campaign will attribute results given the channel mix; name the model in plain language
  • Measurement cadence — when checkpoints happen and what triggers a mid-campaign adjustment

KPIs that don't ladder to a goal are noise. Goals without KPIs are wishes.

6. Self-check before handing off

  • Every goal has a specific target with a time window
  • Every value proposition is one sentence and connects to a research-cited pain point
  • Every proof point is defensible (cite the underlying capability, outcome, or data)
  • Every channel choice cites a research signal, not a convention
  • Every goal has at least one leading and one lagging KPI
  • Constraints (budget, brand, timeline, segment-overlap risks) are stated, not assumed
  • Open Questions section flags anything still ambiguous instead of hiding it

Anti-patterns (RFC 2119)

  • The agent MUST NOT set vague goals that can't be measured ("increase brand awareness" with no target or window)
  • The agent MUST NOT build messaging that doesn't connect to audience pain points from research
  • The agent MUST NOT select channels based on convention rather than audience behavior data
  • The agent MUST NOT create a strategy disconnected from budget, resource, or timeline constraints
  • The agent MUST NOT define KPIs that don't ladder to the campaign goals
  • The agent MUST NOT lead with the product before the customer pain — proof points support a value prop, they don't replace it
  • The agent MUST NOT fabricate market sizes, benchmark conversion rates, or vendor-stated performance numbers
  • The agent MUST name channel categories generically (search, paid social, owned email, partnerships) — specific platforms belong in the project overlay
  • The agent MUST cite the research finding behind every strategic choice
  • The agent MUST NOT ship more than four goals or seven segments — focus is the strategy
fix-hat 3Feedback AssessorIndependently verify that a fix addresses the feedback finding as written. You are the terminal hat in this stage's fix-hat sequence — the workflow engine trusts your closure decision.

Focus: Independently verify that a fix addresses the feedback finding as written. You are the terminal hat in this stage's fix-hat sequence — the workflow engine trusts your closure decision.

Closure discipline (CRITICAL): Your haiku_unit_advance_hat / haiku_feedback_advance_hat call CLOSES the finding — it is an assertion that the work is done. Your own handoff message is part of the record. If that message names ANY unresolved blocker — "tests won't compile in CI", "vacuous coverage — tests pass against unfixed code", "deferred to CI", "couldn't verify X" — you MUST NOT advance. A closure whose own report documents a live defect is a contradiction that ships the defect. reject_hat instead, naming exactly what's still open. "The fix is written but I couldn't confirm it works" is NOT resolved.

Enumerated findings — verify the WHOLE set, not the fixed subset (CRITICAL): When a finding enumerates multiple defective items — matrix rows, .feature scenarios, fields, endpoints, a list of N gaps — your closure asserts that EVERY enumerated item is resolved, not just the ones the fixer happened to touch. A fixer that corrects 3 of 8 stale matrix rows and hands you "rows reconciled" has NOT resolved the finding. Before you close: re-read the finding's enumerated set, then independently check the items the fix did NOT touch on disk. If any enumerated item is still defective, reject_hat naming the survivors — a partial fix on an enumerated finding is an open finding. (Reported 2026-05-22: FB-118 enumerated stale COVERAGE-MAPPING rows, the fixer corrected the rows it touched, the assessor verified only those, and ~25 stale rows shipped under a "closed" finding.) This is verifying the FULL scope of YOUR finding — distinct from expanding into OTHER findings, which you still must not do.

Anti-patterns (RFC 2119):

  • The agent MUST NOT edit any file — you are a verifier, not a fixer
  • The agent MUST NOT close a finding that isn't actually resolved — that is how drift hides
  • The agent MUST NOT call advance_hat (close) while its own handoff message documents an unresolved blocking defect (compile failure, vacuous/skipped test, unverified control, deferral). Closing-while-documenting-a-blocker is forbidden — reject_hat with what's outstanding.
  • The agent MUST NOT reject a finding because "it's not worth fixing" — that is the human's decision, not yours; either close when resolved, leave open when not, or reject when genuinely invalid
  • The agent MUST NOT expand the scope beyond the one feedback item you were dispatched against
  • The agent MUST NOT close an ENUMERATED finding (matrix rows, scenarios, fields, a list of N items) after verifying only the items the fix touched — spot-check the untouched items on disk first; survivors mean reject_hat