Purpose
Turn a business context into a build-ready pricing dashboard specification — metric formulas, chart specs, and panel layout — so a data scientist or BI analyst can implement in Looker, Tableau, Metabase, or a custom app without re-deriving definitions or debating what to track. Solves the gap between "we need a pricing dashboard" and an actual, decision-oriented spec.
When to Use
Invoke this skill when the user asks to:
- Design or spec a pricing performance dashboard
- Define pricing KPIs and their exact formulas (elasticity, price realization, discount leakage, etc.)
- Decide which charts to build for pricing monitoring and how to structure them
- Translate a pricing strategy into a set of tracked metrics
- Set up alerting or thresholds for pricing anomalies
- Standardize pricing metric definitions across a data team
Inputs Required
Before starting, collect:
- Business model: Marketplace / proptech / e-commerce (and whether transactional, subscription, or take-rate based)
- Pricing model: Fixed, dynamic, auction, tiered, take-rate/commission, surge, or usage-based
- Primary pricing decision: What the dashboard should help someone decide (e.g., set floor prices, adjust take rate, optimize discounts)
- Available data: Transaction-level data? Competitor prices? Cost/margin data? Inventory or supply data?
- Audience: Who reads it — pricing manager, exec, or analyst — and cadence (real-time / daily / weekly)
- Key segments: How pricing varies (category, geography, customer tier, listing type, supply cohort)
- North-star pricing goal: Revenue, margin, GMV, take rate, conversion, or fill rate
If pricing model or available data is unspecified, assume transaction-level data exists, dynamic pricing, and note which advanced metrics require additional data feeds.
Instructions
Step 1: Establish the Dashboard Decision Frame
Anchor the spec to the primary pricing decision. State in one line what a viewer should be able to answer within 30 seconds of opening the dashboard, and what action follows. All downstream metric and chart choices serve this frame. Explicitly note the north-star metric that headlines the dashboard.
Step 2: Define 18-25 Metrics Across 5 Layers
For each metric, provide: name, exact formula (using named fields, not prose), grain (row-level, daily, weekly), why it matters for this business, and good/watch/alert direction.
Layer 1 — Revenue & Volume (4-5 metrics) Headline outcomes. Include GMV/net revenue, order/transaction volume, average order value or average transaction price, and (marketplace) take rate = platform revenue / GMV. Use consistent revenue recognition (gross vs. net of refunds).
Layer 2 — Price Realization & Leakage (4-5 metrics) Where intended price differs from captured price. Include price realization = actual price / list-or-target price, discount depth = (list − sold) / list, discount penetration = discounted units / total units, and discount leakage = sum(list − sold) over the period. For take-rate models, include effective take rate vs. posted take rate.
Layer 3 — Elasticity & Conversion (4-5 metrics) How demand responds to price. Include price elasticity of demand = %Δquantity / %Δprice (state the estimation window and controls), conversion rate by price band, sell-through / fill rate, and price-to-conversion curve inputs. Flag elasticity as directional unless experimental or instrumented variation exists.
Layer 4 — Competitive & Positioning (3-4 metrics) Only if competitor data is available; otherwise mark as "requires external feed." Include price index vs. market = own price / competitor median, win rate at price (auction/bid models), and price rank distribution. For proptech, include days-on-market by price-to-estimate ratio.
Layer 5 — Margin & Efficiency (3-4 metrics) Profit quality. Include contribution margin = (price − variable cost) / price, margin per transaction, and (if cost data missing) contribution margin proxy = net revenue after platform costs. Include revenue per available unit/listing where supply is constrained.
Step 3: Specify Charts for Priority Metrics
For each priority metric (aim for 10-14 chart specs), provide a build-ready spec:
- Chart type: line, bar, heatmap, scatter, cohort grid, funnel, distribution/histogram, or KPI tile
- X-axis / Y-axis: exact fields and units
- Series / breakdown: what splits the data (segment, price band, time)
- Aggregation: sum, avg, median, weighted avg — specify weighting field
- Time grain & window: e.g., daily, trailing 90 days
- Reference lines: target, prior period, or threshold overlays
Choose chart types deliberately: elasticity and price-to-conversion → scatter with fitted curve; discount leakage → stacked bar over time; price index → line vs. reference line at 1.0; sell-through by price band → heatmap; conversion funnel by price tier → funnel.
Step 4: Design the Layout by Decision Layer
Specify a top-to-bottom panel layout, grouped so the eye moves from "what happened" to "why" to "what to do":
- Row 1 — KPI header tiles: 4-6 north-star and outcome metrics with period-over-period deltas
- Row 2 — Trend row: primary outcome and price realization over time
- Row 3 — Diagnostic row: elasticity/conversion and leakage charts
- Row 4 — Segment breakdown: the layer that matters most for this business (category, geo, or supply cohort)
- Row 5 — Competitive/margin detail: deeper analysis for analysts State default filters (date range, segment) and which panels are cross-filtered.
Step 5: Define Alert Thresholds
For 5-8 metrics, specify a monitoring rule: metric, condition (e.g., discount leakage > 8% of GMV, price index drifts below 0.95, elasticity estimate crosses −1.0), comparison basis (prior period / rolling baseline), and the pricing action the alert should trigger.
Step 6: List Data Requirements
Enumerate the source tables/fields the spec assumes (transactions, list prices, costs, competitor feeds, inventory). Flag every metric that cannot be computed with the stated available data and name the missing input.
Output Format
- Structure: Six sections following the steps; metrics in a table (name / formula / grain / why / direction); chart specs as structured blocks; layout as a labeled row-by-row diagram in text
- Length: 1,300-1,800 words
- Format: Mixed — metric and threshold tables, structured chart-spec blocks, ordered layout description
- Sections: Decision Frame → Metric Definitions (5 layers) → Chart Specs → Layout → Alert Thresholds → Data Requirements
- Tone: Analytics specialist — precise, formula-driven, implementation-ready; no marketing language
Quality Checklist
Before delivering output, verify:
- [ ] Metric count is 18-25 and distributed across all 5 layers (count them)
- [ ] Every metric has an exact formula using named fields — not a prose description
- [ ] 10-14 chart specs each include chart type, both axes, breakdown, and aggregation
- [ ] Layout assigns every priority metric to a specific row and states default filters
- [ ] 5-8 alert thresholds each name a condition AND the pricing action it triggers
- [ ] Every metric requiring data the user didn't confirm is explicitly flagged with the missing source
Notes
- Formulas must match the pricing model: take rate and effective take rate are central for marketplaces; days-on-market and price-to-estimate ratio for proptech; discount leakage and margin for e-commerce.
- Elasticity from observational data is directional — recommend an A/B or geo test before treating it as a control lever, and label it as such in the spec.
- Weighted averages beat simple averages for price metrics; always specify the weighting field (units or revenue) to avoid Simpson's paradox across segments.
- If the user names a specific BI tool, adapt chart-type vocabulary to that tool's chart library, but keep the metric formulas tool-agnostic.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills