菜单
Audience segmentation toolkit — RFM, behavioral cohorts, JTBD, ICP for B2B, CDP/CRM integration patterns, persona archetype templates.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Five named aesthetic bundles (Ethereal Glass, Editorial Luxury, Soft Structuralism, Minimalist Editorial, Industrial Brutalist) with surface / typography / motion / texture specs and a profile-to-archetype decision matrix.
Canonical CreativeBrief markdown + JSON schema, aligned 1:1 with Hydra CreativeBrief envelope. Sections cover vision, references, aesthetic, technical approach, key shots/messages, production, guardrails, approvals.
HITL gate catalog, audit-log shape, DecisionRecord schema, three-lines-of-defense mapping, v1 gate IDs (brand-consistency, regulated-claims-review, budget-cap, attribution-soundness, ip-clearance).
Brand safety and regulated-claims rulebook — toxicity/bias categories, FDA/FTC/FCA/GDPR rules, prohibited-content registry, brand-guideline enforcement, HITL escalation matrix, verdict schema.
Six industry-specific campaign playbooks — funnel stages, KPI targets, channel mix, cadence, critical gates, failure modes, blueprint examples.
Experimentation toolkit — A/B test design, sample-size from MDE+power, MAB, sequential testing, causal inference (DiD/synth-control/RDD), pitfalls, reporting template.
| name | audience-segmentation |
| description | Audience segmentation toolkit — RFM, behavioral cohorts, JTBD, ICP for B2B, CDP/CRM integration patterns, persona archetype templates. |
| user-invocable | true |
| allowed-tools | ["Read","Glob","Grep","Write"] |
Owned by audience-persona; consumed by campaign-strategist, lifecycle-crm, contextual-copywriter, and media-buyer-bidder. Four lenses: RFM (transactional), Behavioral cohorts (event-based), JTBD (motivational), ICP (B2B firmographic). Use multiple lenses in combination — no single lens is sufficient.
For each customer, compute:
Bin each into 1-5 quintiles (5 = best). Concatenate to RFM score (e.g. "545").
| Segment | RFM pattern | Action |
|---|---|---|
| Champions | 4-5, 4-5, 4-5 | Loyalty, referrals, early access |
| Loyal | 3-5, 4-5, 3-5 | Upsell, advocacy ask |
| Potential loyalist | 4-5, 1-2, 1-3 | Onboarding deepening, second purchase |
| New customers | 4-5, 1, 1 | Welcome series, expectation setting |
| Promising | 3-4, 1, 1-2 | Activation triggers |
| Needs attention | 2-3, 2-3, 2-3 | Targeted re-engagement |
| About to sleep | 2-3, 1-2, 1-2 | Re-engagement campaign |
| At risk | 1-2, 2-5, 2-5 | Winback before churn |
| Can't lose them | 1, 4-5, 4-5 | High-touch, possible HITL |
| Hibernating | 1-2, 1-2, 1-2 | Low-cost reactivation |
| Lost | 1, 1, 1 | Suppress or sunset |
| Industry | R quintile window | F window |
|---|---|---|
| DTC fast-moving (beauty, food) | 30/60/90/180/365 days | 365 days |
| DTC durable (apparel, home) | 60/120/180/365/730 days | 730 days |
| B2B SaaS (subscription) | Use active/churn lifecycle instead | |
| Pro Services | 180/365/730/1095 days | 1095 days |
Define cohorts by event sequence, not demographics. Standard funnel:
signed_up -> activated -> engaged -> power_user -> at_risk -> churned
| State | Trigger | Typical threshold |
|---|---|---|
| signed_up | account created | t = 0 |
| activated | first meaningful action | within D1-D3, varies by product |
| engaged | recurring meaningful action | >=3 sessions in 7 days |
| power_user | top-decile usage | top 10% by frequency or depth |
| at_risk | usage decay | 50% drop in 14-day rolling avg |
| dormant | no activity | no event in 30 days |
| churned | cancellation or 90-day dormant | explicit cancel or no event 90d |
| Day | D0 | D1 | D7 | D14 | D30 | D60 | D90 |
|---|---|---|---|---|---|---|---|
| % active | 100% | x | x | x | x | x | x |
Compare cohorts week-over-week to detect activation regressions.
Frame the customer's underlying motivation, not features.
When [situation],
I want to [motivation / desired progress],
So I can [desired outcome / emotional or social payoff].
Example (B2B observability SaaS):
When a production incident wakes my on-call at 3am, I want to find the root cause in under 10 minutes, so I can restore service before customers notice and protect my team's morale.
Firmographic + technographic + behavioral.
icp:
firmographic:
industry: ["SaaS", "Fintech"]
revenue_band: "$50M - $500M ARR"
employee_count: "200 - 2000"
region: ["North America", "EMEA"]
funding_stage: "Series B - D"
technographic:
must_have: ["Kubernetes", "AWS or GCP"]
integrations_used: ["PagerDuty", "Slack"]
excludes: ["mainframe-only", "fully on-prem"]
behavioral:
growth_signals: ["hiring SREs", "recent funding"]
triggers: ["recent outage news", "platform team posted"]
buying_committee:
champion: "Sr. SRE / Platform Eng Manager"
economic_buyer: "VP Engineering / CTO"
technical_buyer: "Lead SRE"
blocker: "Security / Compliance"
disqualifiers:
- "Hard requirement for on-prem"
- "Annual contract value < $25k"
Score = w1firmographic_match + w2technographic_match + w3*behavioral_signal. Use 0/0.5/1 per field. Score >= 0.7 = Tier-1 ICP.
| System | Family | Marketing role | Notes |
|---|---|---|---|
| Segment | CDP | Source-of-truth event router | Best for product-led data |
| mParticle | CDP | Enterprise event router with consent mgmt | Strong for regulated |
| RudderStack | CDP | Open-source / warehouse-native | Cost-effective at scale |
| HubSpot | CRM | Mid-market sales+marketing | Native marketing automation |
| Salesforce + Pardot/MCAE | CRM | Enterprise B2B | Powerful but heavy |
| Klaviyo | ESP/CRM | DTC ecommerce | Best-in-class for DTC lifecycle |
| Braze | Cross-channel | Cross-channel orchestration | Mobile-first lifecycle |
| Customer.io | Cross-channel | Mid-market event-based | Strong for SaaS |
{
"userId": "u_123",
"anonymousId": "a_456",
"event": "Product Viewed",
"properties": {
"sku": "...",
"category": "...",
"price": 39.99
},
"context": {
"campaign": {"name": "...", "source": "...", "medium": "..."},
"consent": {"marketing": true, "analytics": true}
},
"timestamp": "2026-05-19T12:34:56Z"
}
Consent flag must be carried on every event — required for GDPR/CCPA.
Three example personas across industries (substitute per profile).
marketing-attribution (cohorts feed MTA)lifecycle-marketing (cohorts drive cadence)experimentation-design (cohorts are randomization units)