| name | instrumentation-spec-data |
| description | This skill writes the instrumentation specification defining what events and properties to track. Use when asked to design an event taxonomy, define tracking properties, or create a measurement plan. Also consider when a new feature enters development without a tracking spec. Suggest when a PRD has success metrics but no corresponding event definitions.
|
| department | data-growth |
| agent | analytics-lead |
| version | 1.0.0 |
| complexity | complex |
| related-skills | ["instrumentation-clarity-reviewer","instrumentation-planner-data","instrumentation-implementer-data","goal-framer-data"] |
| triggers | ["write instrumentation spec","define tracking spec","analytics tracking specification","data instrumentation plan","event taxonomy spec"] |
instrumentation-spec-data
Agent: Analytics Lead
L1 analytics leader (1x) responsible for search demand validation, market sizing, goal framing, instrumentation strategy, and north star metric governance.
Department ethos: ideal-data-growth.md
Skill Description
The instrumentation spec writer produces the authoritative document defining every event name, trigger condition, property schema, and data type required to measure a product surface, ensuring that analysts can construct any required funnel, cohort, or retention query from the captured data.
When to Use
- When a new product feature enters development and needs event tracking defined before implementation.
- When a PRD contains success metrics that require new or modified instrumentation.
- When an existing product surface has ad-hoc tracking that needs formalization into a consistent taxonomy.
- When a platform migration requires re-specifying the event schema for a new analytics tool.
Workflow
- Gather measurement requirements: Extract every metric, funnel step, and analytics question from the PRD, goal framework, and stakeholder interviews. List the queries the data must support.
- Map user journeys: Document each user journey through the product surface step by step. Identify the action or state change at each step that constitutes a trackable event.
- Define event taxonomy: Name each event using the object_action convention (e.g.,
checkout_started, item_added). Group events by journey phase. Assign a unique event ID.
- Specify properties per event: For each event, define every property with its name, data type, allowed values (enum or range), nullability, and description. Include standard context properties (user_id, session_id, timestamp, platform, app_version).
- Map events to metrics: Create a traceability matrix linking each analytics goal to the events and properties required to compute it. Identify any goal that lacks full coverage and add events to close the gap.
- Define data quality rules: Specify validation rules per event — required properties, value ranges, expected volume per hour. These rules feed into instrumentation verification.
- Document privacy classification: Tag each property with a privacy tier (anonymous, pseudonymous, PII). Flag properties requiring consent gating or suppression in certain jurisdictions.
- Review and publish: Submit the spec for clarity review. Incorporate feedback. Publish the approved spec as the source of truth for implementation.
Anti-Patterns
- Tracking everything: Specifying events for every micro-interaction without tying them to an analytics question creates storage cost and query noise. Why: unused events dilute the signal-to-noise ratio and increase pipeline maintenance burden.
- Vague property definitions: Defining a property as "value: string" when it holds a numeric amount causes downstream type-casting failures and aggregation errors. Why: loose typing breaks silently — dashboards show null or NaN instead of erroring visibly.
- No traceability matrix: Writing events without mapping them to the metrics they support makes it impossible to verify completeness. Why: you discover coverage gaps only when an analyst tries to build a query and finds the data missing.
- Ignoring server-side events: Specifying only client-side events misses backend actions (payment processing, subscription changes, API calls). Why: critical business events often happen server-side and are invisible to client tracking.
- Skipping privacy classification: Omitting privacy tiers from the spec risks shipping PII-laden events that violate GDPR or CCPA. Why: retroactive PII scrubbing from event stores is expensive and often incomplete.
Output
Success:
- A complete instrumentation spec document containing event taxonomy, property schemas with types and constraints, traceability matrix, data quality rules, and privacy classifications.
- Stakeholder and clarity-reviewer approval confirming the spec is implementation-ready.
Failure:
- The spec ships with coverage gaps that block metric computation. Report which metrics are uncomputable, which events are missing, and the revised spec required.
- Properties are mis-typed or under-constrained, causing data quality issues in production. Report the affected events and corrected definitions.
Related Skills