pmm-metrics

PMM Metrics

How This Works

Picks one North Star Metric that predicts customer value and revenue, names the 3-5 Input Metrics that move it, then rounds out a scorecard across four categories (Financial, Customer, Product & GTM, Process & Growth), capped at 2-4 metrics per category — 8-16 total, small enough that a team actually updates it every 1-2 weeks instead of abandoning it by Q2.

Every target and current value comes from the user — this skill doesn't invent numbers. If brain Section 2 (ICP/Beachhead) is loaded, it uses that to check whether a metric is a real leading indicator for the confirmed segment or just generic. Output is chat-only, disposable — the brain has no Metrics section to write to (all 6 existing sections are claimed), same treatment experiment-ideas (pmm-growth) gives its output.

Step 0 — Load brain Sections 1 (Product), 2 (ICP/Beachhead) if present, and guardrails from /context/meta-patterns.md.

Step 1 — Classify the business game: Attention, Transaction, or Productivity.

Step 2 — Define the North Star Metric and validate it against 7 criteria.

Step 3 — Define 3-5 Input Metrics that most directly move the North Star.

Step 4 — Build the supporting scorecard: 2-4 metrics per category across Financial, Customer, Product & GTM, and Process & Growth, each with a target, weight, and cadence.

Step 5 — Self-check the full set against sprawl and vanity-metric traps.

Step 6 — Deliver the scorecard as a markdown table the user saves themselves — this skill has no durable output file of its own.

Step 7 — Learning Close: log the session to /context/skill-sessions.md.

Trigger

• When: Choosing a North Star Metric, building a measurement

framework from scratch, auditing an existing metrics list that's sprawled past what anyone tracks, or preparing a scorecard for leadership/exec reporting.

• Not for: Quarterly OKRs once metrics exist → pmm-okrs

(pmm-execution) expresses target change in metrics this skill defines; run this skill first if no North Star exists yet. Launch tier assignment → go-to-market-strategy. Pressure-testing one experiment's hypothesis → experiment-doc (pmm-execution) — feed it this skill's metric as the guardrail, don't invent a new one. Post-launch metrics review → retro (pmm-execution) looks backward at what this skill's scorecard already tracks.

• Example prompts:
• "What should we be measuring?"
• "Help me define our North Star metric"
• "Build a PMM scorecard for leadership"
• "What metrics should Product Marketing track?"
• "Our metrics list has 25 things on it and nobody updates it"
• "Set up a measurement framework for Q1"

Inputs

• Args: Business model description (how revenue is made, what

customers do in-product), current metrics being tracked (if any), reporting cadence and audience (team-level vs. exec-level).

• Defaults: No brain and no business context given → ask directly

(Step 1). Brain exists → load Section 1/2 silently, ask only for what isn't already on file.

• Context keys:
• /foundation/brain.md — optional. Section 1 (Product) informs

business-game classification; Section 2 (ICP/Beachhead) informs which Customer and Product & GTM metrics are actual leading indicators for the confirmed segment vs. generic.

• /context/meta-patterns.md — optional; guardrails from prior

metrics sessions.

• Brain contract: Reads Sections 1, 2. Writes nothing — the brain's

sections are all claimed; adding a Metrics section is a product-marketing-context schema change, out of scope here. Scorecard is chat-only output, same as experiment-ideas.

Connectors (Optional)

No connectors? Paste the material and I'll work from that. The output has the same shape. Pulled facts are drafts: each is tagged with category, tool and date, and nothing is written until you confirm.

Pre-flight

• Load /foundation/brain.md Sections 1, 2 if it exists — silently.
• Load /context/meta-patterns.md if it exists, and surface any

guardrail that has fired 2+ times in prior metrics sessions.

• If no brain found: surface once, non-blocking — "No brain found. I'll

ask directly about your business model and customer segment instead. Metrics will be less sharply targeted without ICP/beachhead context — run product-marketing-context first for a tighter fit, or continue now."

• No hard block. This skill runs entirely from user-stated business

context; brain absence degrades precision, not capability.

Steps

Step 1 — Classify the Business Game

Ask, or infer from brain Section 1 if loaded:

"How do customers primarily get value from your product — by spending time in it, by transacting through it, or by getting something done more efficiently with it?"

Classify into exactly one:

State the classification with one sentence of rationale — never assume silently. Most B2B SaaS lands in Productivity; treat that as a hypothesis to confirm, not a default to skip past.

Step 2 — Define the North Star Metric

A North Star Metric (NSM) is not multiple metrics, a revenue/LTV number, an OKR, or a strategy. It is a single, customer-centric KPI reflecting the value customers get from the product — a leading indicator of long-term business success.

Propose one candidate matched to Step 1's business game, then validate it against all seven criteria before confirming:

1. Easy to Understand — clear enough that everyone in the org

comprehends it without a footnote.

1. Customer-Centric — reflects value delivered to customers, not

just revenue or internal activity.

1. Sustainable Value — indicates habits and long-term engagement,

not a one-time spike.

1. Vision Alignment — represents meaningful progress toward the

company's stated vision or mission.

1. Quantitative — measurable with clear, numeric tracking.
2. Actionable — teams can directly influence it through product,

marketing, and operational changes.

1. Leading Indicator — predicts future business success and revenue

growth, rather than lagging behind it.

Score each criterion Pass/Fail with one line of evidence. 2+ fails → reject and propose an alternative, never deliver a failing NSM with caveats. Revenue and ARR usually fail #2 (Customer-Centric); NPS often fails #7 (Leading Indicator) without a longitudinal retention/expansion link. Name the specific failure — not a generic "that's not customer-centric."

Step 3 — Define Input Metrics

Define 3-5 Input Metrics — leading indicators that most directly move the North Star. Each must:

• Move more easily, short-term, than the NSM itself
• Have a stated causal link to the NSM, not just topical proximity
• Point to where optimization effort should concentrate

For each: Metric, Why it drives the NSM (causal, one sentence), Who owns moving it (PMM, Product, Sales, CS).

If brain Section 2 (ICP/Beachhead) is loaded, check each against the confirmed segment — generic enough to apply to any customer? Flag it [GENERIC — not segment-specific] rather than including it silently.

Step 4 — Build the Supporting Scorecard

Select 2-4 metrics per category, 8-16 total — past that, the scorecard stops getting updated by Q2. Reference list below, not a requirement to use all of it:

Financial ARR/Revenue/Bookings · Qualified pipeline ($ and count, by stage) · Pipeline sourced from PMM-influenced campaigns · Win rate · Average deal size · Time to close · Expansion revenue (upsell/cross-sell share) · Product-level revenue · Market penetration in target segment

Customer NPS / CSAT · New customer reviews (count + score, e.g. G2/Capterra) · Customer advocacy (case studies, reference customers) · Engagement in sales/customer calls · DAU/MAU (post-launch or feature-specific)

Product & GTM Messaging & positioning effectiveness (win/loss feedback, message testing) · Sales enablement efficacy (% of calls using PMM content) · Content & thought leadership engagement · Competitive edge (battlecard usage, win rate vs. named competitor) · Market leadership (analyst recognition) · Feature/product adoption rate · New use case discovery

Process & Growth GTM planning and MRD/PRD creation velocity · Feedback loop efficiency with GTM team · Asset utilization (content actually used vs. produced) · Funnel conversion rate at each stage (MQL→SQL, SQL→Opp, Opp→Closed) · Pricing & packaging effectiveness

For each metric: Name, plain-language Definition, Target, Weight (sums to 100% across the full scorecard, not per category), Cadence (default biweekly).

Present as a table, grouped by category, with a subtotal weight shown per category.

Step 5 — Self-Check Against Sprawl and Vanity Metrics

Before delivering, check the full set:

• ≤16 total metrics. Over that, cut — ask which 2 matter least,

don't just append.

• Every metric has a named owner. No owner, no metric.
• No duplicate metrics worded differently (e.g. "content engagement"

vs. "thought leadership engagement") — merge or cut.

• Every category has ≥1 metric — zero means Step 4 was skipped for

it; flag and revisit.

• Weights sum to exactly 100%.
• Flag anything trackable but not actionable by a named function —

[VANITY — no owner can move this].

Step 6 — Deliver the Scorecard

Present the finished set — NSM, Input Metrics, and categorized scorecard table — as the session's final chat output. State plainly that this is disposable: "Copy this into your tracker of choice (the Product Marketing Scorecard template works well for ongoing tracking) — this skill doesn't save it for you."

Step 7 — Learning Close

End every completed session by appending one entry to /context/skill-sessions.md, in the format defined in product-marketing-context/.claude-plugin/skill-sessions-format.md (Type A — execution session):

type: execution
skill: pmm-metrics
session_date: [YYYY-MM-DD]
pattern: [one falsifiable statement about what happened this session, or "none"]
source: [surprised / wrong / missing / N/A]

Write this row directly — do not ask the user for permission. This is the only durable write this skill makes — see Step 6: the scorecard itself is chat-only, disposable output. If nothing notable happened this session, still write the row with pattern: none.

Outputs

• Files written: /context/skill-sessions.md — one row per session

(Step 7). Nothing else — see Step 6.

• Chat output format: Business game → NSM with 7-criteria scoring →

Input Metrics with causal rationale and owners → categorized scorecard table → self-check results.

• External side effects: N/A
• Next skill: After delivery, check

product-marketing-context/.claude-plugin/next-skill-map.md for "After pmm-metrics" and surface it as a question — never auto-run.

Verification

• Business game classified with rationale, not assumed.
• NSM scored against all 7 criteria individually, not aggregated.
• 3-5 Input Metrics, each with causal link and named owner.
• Scorecard: 8-16 metrics across all 4 categories, each with target,

weight, cadence.

• Weights sum to exactly 100%.
• Sprawl/vanity self-check run and shown.
• Scorecard delivered with the explicit "not saved for you" note (Step 6).
• Session logged (Step 7).

Commands

/classify

Run Step 1 only — classify the business game and stop. Useful when the user wants the framework concept without building a full scorecard yet.

/classify

/north-star

Run Steps 1-3 only — business game, NSM, and Input Metrics. Skips the broader scorecard.

/north-star

/scorecard

Run Step 4 only, assuming NSM and Input Metrics are already defined (stated inline in this session). Builds the supporting scorecard directly.

/scorecard

/audit

Pressure-test an existing metrics list the user provides — apply Step 5 sprawl/vanity self-check against it directly, without redefining the NSM.

/audit

Operating Rules

• NSM is singular. Never present more than one as interchangeable

options — pick one, name why the runner-up lost.

• Revenue-shaped metrics fail customer-centric by default. Challenge

any NSM that's really a revenue or company-health metric in disguise.

• Every metric needs a named owner. Unowned is a chart, not a

management tool.

• Cap the scorecard at 16. Cut before adding.
• Weights sum to 100%, shown explicitly. Never leave weighting as an

exercise for the user.

• State output is disposable. No brain section to write to — never

imply otherwise.

• Business game is stated, never assumed — even the obvious cases.
• Flag generic and vanity metrics visibly — never quietly drop or

quietly include them.

Quality Gate

Do Not Use For

• pmm-okrs (pmm-execution) — quarterly OKRs once metrics exist; this

skill defines what to measure, not the quarter's target change.

• go-to-market-strategy — launch tier and channel strategy; feed it

this skill's metrics as success-metric inputs, don't substitute.

• experiment-doc (pmm-execution) — pressure-testing one hypothesis;

its guardrail metric should come from this skill's scorecard.

• retro (pmm-execution) — reviewing how metrics performed after a

launch; this skill defines the framework, retro looks backward at it.