What are the four layers of a measurement strategy?

A north star and KPI hierarchy, an event taxonomy, dashboards and reports, and attribution and segmentation. Each feeds the next: the north star and KPIs define what matters, the taxonomy captures it, the dashboards present it for decisions, and attribution connects cause to effect. A strategy missing a layer tends to produce data that cannot answer the questions the team actually has.

What makes a good north star metric?

One metric, singular, that captures customer-perceived value and leads to revenue without being revenue itself, because revenue sits too far downstream to steer day-to-day work. Examples include weekly active users, completed jobs, revenue-generating sessions, and hours of value delivered. Underneath it sits the AARRR hierarchy (acquisition, activation, engagement, retention, monetization), which works because it covers the full lifecycle rather than a single moment.

How should events be named?

Verb plus noun, past tense, in snake_case: signed_up, created_project, completed_checkout. Fire one event per discrete action rather than a vague 'interacted_with_modal', attach properties for context (signup_method, referrer, plan), and standardize property names so user_id means user_id everywhere. Document everything in a tracking plan, because a plan that lives in someone's head is one no one follows, and inconsistent naming turns every report into detective work.

Which attribution model should I use?

Pick one primary model and use others for validation, because all models lie and some lie usefully. First-touch credits the first interaction (useful for awareness), last-touch credits the final one (the default in many tools, and often misleading), linear spreads credit evenly, time-decay weights recent touches, position-based weights the first and last, and data-driven models use machine learning. Treating any single model as the truth is a failure mode; the differences between models are themselves informative.

How is this different from product-analytics-setup?

Analytics strategy is the strategic measurement plan: the north star, the KPI hierarchy, the event taxonomy design, the dashboard architecture, and the attribution approach. Product-analytics-setup is the execution: instrumenting those events correctly the first time, with the property design, naming, and schema discipline that keeps the data trustworthy. Plan the measurement here; implement it there. The two are companions, strategy and execution of the same system.

Skill · Analytics strategy

Analytics strategy.

Measurement that drives decisions, not vanity dashboards.

Design a measurement framework that produces decisions. Four layers cover it: a north star and KPI hierarchy, an event taxonomy, dashboards and reports, and attribution and segmentation. Each layer feeds the next, and the whole thing works backward from the questions the team needs to answer.

The output is a tracking plan: a living document that names the north star, every event, the naming conventions, the privacy rules, and the governance. This skill plans the measurement; instrumenting it is a separate job.

Audience: founders, PMs, and growth teams setting up analytics, auditing an existing setup, or turning a pile of data into a way to make decisions.

View the skill on GitHub Browse the full catalog

The framework

Four layers, each feeding the next.

A complete measurement strategy covers all four. Start from the questions the team needs to answer, then work down.

01North star and KPI hierarchy: one metric that captures customer-perceived value, with the AARRR hierarchy underneath it: acquisition, activation, engagement, retention, monetization.
02Event taxonomy: the vocabulary for what users do. Verb plus noun, past tense, snake_case; one event per action; properties for context; standardized names; documented.
03Dashboards and reports: one audience per dashboard, one question per chart, annotations and comparisons for context, leading with the decision the dashboard supports.
04Attribution and segmentation: one primary attribution model validated by others, and segments chosen by what causes different behavior rather than what is easy to track.

The output

Eight parts of the tracking plan.

The strategy lands as a tracking plan, a living document that keeps measurement consistent as the product grows.

01
Goals and KPIs
The business objectives, the north star, and the KPI hierarchy underneath it.
02
Event catalog
Every event, with its properties, when it fires, why it is tracked, and who owns it.
03
User properties
Persistent attributes carried across events: plan, signup date, role.
04
Page taxonomy
Page categories and the properties tracked consistently across page views.
05
Naming conventions
Snake_case, verb plus noun, and standardized property keys, so reports are not detective work.
06
Implementation notes
Client-side versus server-side, SDK details, and sampling decisions.
07
Privacy and compliance
PII rules, consent handling, and data retention, kept out of event properties.
08
Governance
Who can add events, the review process, and the change log that keeps the plan trustworthy.

Pairs with these platforms

Where the events and dashboards live.

The strategy is tool-agnostic, but the event taxonomy and the dashboards live in a product analytics platform. Mixpanel and Amplitude are dedicated product analytics suites; PostHog pairs analytics with experimentation in one place.

Reference files

The reference that goes alongside the SKILL.md.

references/event-taxonomy-template.md
A starter event catalog with naming patterns for common product types.

Browse all reference files on GitHub

Bridges to other skills

What executes the plan, and what reads it.

Analytics strategy is the plan. These cover the instrumentation, the testing, and the qualitative side.

Execution
product-analytics-setup
Instruments the events correctly the first time: the execution to this skill's strategy. Plan the taxonomy here, then implement it there.
Acts on the data
cro-optimization
Conversion testing runs on the tracking and funnel data this skill defines. The strategy makes the funnel measurable; CRO tests changes to it.
The why
ux-research
Analytics shows what users do and where they drop; research explains why. The two are different instruments pointed at the same product.
SEO measurement
seo-onpage
Search performance has its own metrics and tools. Reach for the SEO skills for organic measurement rather than the product analytics plan.

Open source under MIT

Read the SKILL.md on GitHub.

The skill source lives in the rampstackco/claude-skills repository alongside dozens of other skills covering the full lifecycle of brand and product work. This page is a structured overview; the SKILL.md is the source. MIT licensed.

View SKILL.md Browse the full catalog

Frequently asked questions.

What are the four layers of a measurement strategy?: A north star and KPI hierarchy, an event taxonomy, dashboards and reports, and attribution and segmentation. Each feeds the next: the north star and KPIs define what matters, the taxonomy captures it, the dashboards present it for decisions, and attribution connects cause to effect. A strategy missing a layer tends to produce data that cannot answer the questions the team actually has.
What makes a good north star metric?: One metric, singular, that captures customer-perceived value and leads to revenue without being revenue itself, because revenue sits too far downstream to steer day-to-day work. Examples include weekly active users, completed jobs, revenue-generating sessions, and hours of value delivered. Underneath it sits the AARRR hierarchy (acquisition, activation, engagement, retention, monetization), which works because it covers the full lifecycle rather than a single moment.
How should events be named?: Verb plus noun, past tense, in snake_case: signed_up, created_project, completed_checkout. Fire one event per discrete action rather than a vague 'interacted_with_modal', attach properties for context (signup_method, referrer, plan), and standardize property names so user_id means user_id everywhere. Document everything in a tracking plan, because a plan that lives in someone's head is one no one follows, and inconsistent naming turns every report into detective work.
Which attribution model should I use?: Pick one primary model and use others for validation, because all models lie and some lie usefully. First-touch credits the first interaction (useful for awareness), last-touch credits the final one (the default in many tools, and often misleading), linear spreads credit evenly, time-decay weights recent touches, position-based weights the first and last, and data-driven models use machine learning. Treating any single model as the truth is a failure mode; the differences between models are themselves informative.
How is this different from product-analytics-setup?: Analytics strategy is the strategic measurement plan: the north star, the KPI hierarchy, the event taxonomy design, the dashboard architecture, and the attribution approach. Product-analytics-setup is the execution: instrumenting those events correctly the first time, with the property design, naming, and schema discipline that keeps the data trustworthy. Plan the measurement here; implement it there. The two are companions, strategy and execution of the same system.

Analytics strategy.

Four layers, each feeding the next.

Eight parts of the tracking plan.

Goals and KPIs

Event catalog

User properties

Page taxonomy

Naming conventions

Implementation notes

Privacy and compliance

Governance

Where the events and dashboards live.

Mixpanel

Amplitude

PostHog

The reference that goes alongside the SKILL.md.

What executes the plan, and what reads it.

Read the SKILL.md on GitHub.

Frequently asked questions.