product-analytics

Overview of product-analytics skill

What product-analytics is for

The product-analytics skill helps an AI agent define product KPIs, choose metric frameworks, design dashboards, and interpret retention, cohort, funnel, and feature adoption data. It is built for product managers, growth teams, founders, analysts, and AI-assisted product teams that need structured metric thinking rather than a one-off chart suggestion.

Use it when you are asking questions like: “What should we measure at this product stage?”, “Is this feature actually adopted?”, “Which retention view should go on the dashboard?”, or “How do we turn a vague North Star goal into measurable input metrics?”

Best-fit product analytics jobs

This product-analytics skill is strongest for Product Management workflows where the problem is partly analytical and partly decision-oriented. Good fits include:

• Selecting between AARRR, North Star, and HEART frameworks
• Defining KPIs for pre-PMF, growth, or mature products
• Creating executive, product health, or feature adoption dashboard structures
• Planning cohort retention analysis around signup, activation, or feature exposure
• Interpreting metric movement against product launches and lifecycle stage

It is less useful if you only need SQL generation, warehouse modeling, or a BI-tool-specific setup. The skill provides product analytics reasoning and templates, not a full analytics engineering stack.

What makes this skill practical

The repository includes more than a single prompt file. The core SKILL.md explains when to use the skill and the workflow. references/metrics-frameworks.md gives usable framework guidance for AARRR, North Star, and HEART. references/dashboard-templates.md provides dashboard layouts for executive, product health, and feature adoption views. scripts/metrics_calculator.py adds a lightweight command-line helper for retention, cohort, and funnel calculations from CSV data.

That combination makes the product-analytics skill useful for both planning and first-pass analysis.

How to Use product-analytics skill

product-analytics install and first files to read

To install from the GitHub repository, use your skill manager’s GitHub install flow. For example, if your environment supports npx skills add, the practical command is:

npx skills add alirezarezvani/claude-skills --skill product-analytics

After install, read these files in order:

1. SKILL.md — scope, workflow, KPI guidance, and interpretation rules
2. references/metrics-frameworks.md — AARRR, North Star, HEART, and Goals-Signals-Metrics
3. references/dashboard-templates.md — dashboard structures and KPI blocks
4. scripts/metrics_calculator.py — optional CSV-based calculations for retention, cohorts, and funnels

This reading path matters because the skill is framework-driven. If you skip the references, the agent may produce generic KPI lists instead of stage-appropriate analytics guidance.

Inputs that make product-analytics usage better

For strong product-analytics usage, give the agent product context, stage, user segments, and the decision you need to make. Weak prompt:

Help me create product metrics.

Better prompt:

Use the product-analytics skill. We are a B2B SaaS product in growth stage. Users sign up, invite teammates, create a project, and publish reports. Our suspected activation event is “created first project with at least one teammate.” We need a product health dashboard for the PM and leadership team. Define a North Star candidate, input metrics, activation and retention KPIs, and dashboard layers. Call out missing data and risks.

For retention or funnel work, include the event names, cohort basis, time window, and segments. Example: signup cohort vs first-feature-use cohort will answer different questions.

Suggested workflow for real analysis

Start by asking the skill to choose or compare frameworks, not to jump directly into metrics. A useful sequence is:

1. Define product stage and business model
2. Select the metric framework: AARRR for growth funnels, North Star for strategic alignment, HEART for UX quality
3. Identify the first value moment and activation event
4. Build metric hierarchy: North Star, input metrics, guardrails, diagnostic metrics
5. Design dashboard layers for the audience
6. Run or request cohort, retention, funnel, or feature adoption analysis
7. Translate metric movement into decisions, experiments, or instrumentation gaps

If you have CSV exports, inspect scripts/metrics_calculator.py before asking the agent to compute retention or funnel conversion. The script expects clear user, cohort, activity, and funnel columns; messy event logs may need preprocessing.

product-analytics skill FAQ

Is product-analytics for Product Management or data science?

The product-analytics skill is primarily for Product Management, product strategy, and analytics planning. It helps define what to measure, why it matters, and how to interpret movement. It can support analyst workflows, especially around cohort and funnel framing, but it does not replace a warehouse model, experimentation platform, or statistical notebook.

For product managers, the biggest benefit is turning vague goals into stage-aware KPIs and dashboard requirements that analysts or BI teams can implement.

How is this better than an ordinary analytics prompt?

A generic prompt often returns a broad list of metrics: DAU, MAU, retention, conversion, churn, revenue. This skill gives the agent a more opinionated product analytics workflow: framework selection, stage-specific KPI guidance, dashboard layering, cohort comparison, and feature adoption interpretation.

The included references also reduce ambiguity. Instead of inventing a dashboard from scratch, the agent can use executive, product health, and feature adoption templates as starting points.

When should I not use this skill?

Do not use product-analytics as the main tool when your problem is purely technical, such as writing production SQL, debugging tracking SDKs, designing dbt models, or configuring Amplitude, Mixpanel, Looker, or GA4. It can help specify the metrics and events those tools need, but it is not a vendor implementation guide.

Also avoid using it when you have no product context. Without lifecycle stage, user journey, key events, or business goal, the output will be high-level and less actionable.

How to Improve product-analytics skill

Improve product-analytics results with stronger context

The most important upgrade is better input. Include:

• Product type: SaaS, marketplace, consumer app, internal tool, content product
• Stage: pre-PMF, growth, mature, turnaround, launch
• Core user journey: signup, onboarding, value moment, repeat behavior
• Business model: subscription, usage-based, ads, transaction fee, enterprise sales
• Current concern: activation, retention, monetization, adoption, quality, churn
• Available data: event logs, CRM fields, billing data, surveys, support tickets

This lets the product-analytics skill avoid mismatched frameworks. For example, HEART may fit a UX quality problem, while AARRR is better for acquisition-to-revenue funnel diagnosis.

Watch for common failure modes

Common weak outputs include too many KPIs, vanity metrics without decisions, dashboard designs with no owner, and retention analysis based on a single snapshot. Push the agent to separate:

• Executive metrics from diagnostic metrics
• Leading indicators from lagging outcomes
• Segment-level signal from blended averages
• Feature exposure from true feature adoption
• First use from repeat or sustained usage

A good product analytics answer should tell you what decision each metric supports. If a metric has no owner, threshold, or action path, ask the agent to revise.

Iterate after the first output

After the first answer, improve it with targeted follow-ups:

• “Reduce this to 5 executive metrics and 10 diagnostic metrics.”
• “Rewrite for a pre-PMF product with low traffic.”
• “Add instrumentation events needed to calculate each KPI.”
• “Separate dashboard views for PM, leadership, and growth team.”
• “Identify which metrics are guardrails versus success metrics.”
• “Turn this into an experiment readout template.”

For data-backed work, combine the framework output with the repository’s scripts/metrics_calculator.py where appropriate, then ask the skill to interpret the results in context rather than merely restating percentages.