Illustrative pattern · synthetic data
Platform demo · Amazon Redshift
Warehouse design: dist/sort keys and Spectrum notes
Walk dist/sort recommendations on synthetic warehouse objects, WLM queue metaphor, and Spectrum external toggles that finance can actually consume.
How this maps to Amazon Redshift
Amazon Redshift consulting and warehouse modernization.
- Dist/sort key cards on fact tables
- WLM queue metaphor for BI vs ETL
- Spectrum external object toggle
- Synthetic facts; not a live Redshift cluster
Related consulting page → · All platform demos · Book My BI Diagnostic
Redshift warehouse design
Dist/sort recommendations and Spectrum-style external objects.
Illustrative pattern · synthetic data
Stack
- S3 + Spectrum
- Staging schema
- Dist/sort-keyed facts
- WLM queues
- BI tool
Illustrative pattern · synthetic data
KEY distributed
Illustrative pattern · synthetic data
External / Spectrum
Illustrative pattern · synthetic data
Illustrative pattern · synthetic data
Estimated size by warehouse object (bytes)
Illustrative pattern · synthetic data
KPI recipe: gross margin % Redshift SQL
SELECT d.year_month,
SUM(f.gross_margin)::DECIMAL(18, 4)
/ NULLIF(SUM(f.revenue), 0) AS gross_margin_pct
FROM analytics.fact_invoices f
JOIN analytics.dim_date d ON d.date_key = f.date_key
GROUP BY d.year_month;
Illustrative pattern · synthetic data
Semantic objects
| Object | Type | Notes |
|---|---|---|
analytics.bridge_crm_erp_keys | Redshift table | bridge |
analytics.dim_channel | Redshift table | dimension |
analytics.dim_customer | Redshift table | dimension |
analytics.dim_date | Redshift table | dimension |
analytics.dim_product | Redshift table | dimension |
analytics.dim_region | Redshift table | dimension |
analytics.dim_sales_rep | Redshift table | dimension |
analytics.fact_inventory_backlog | Redshift table | fact |
analytics.fact_invoices | Redshift table | fact |
analytics.fact_opportunities | Redshift table | fact |
analytics.fact_opportunity_stage_hist | Redshift table | fact |
analytics.fact_sales_orders | Redshift table | fact |
analytics.fact_samples | Redshift table | fact |
Total Revenue | View metric | certified · Finance |
Revenue YTD | View metric | certified · Finance |
Gross Margin | View metric | certified · Finance |
Gross Margin % | View metric | certified · Finance |
Open Backlog | View metric | certified · Operations |
Pipeline Value | View metric | certified · Sales |
Weighted Pipeline | View metric | certified · Sales |
Sample Conversion % | View metric | certified · Sales |
Draft Channel Mix | View metric | draft · Sandbox |
Uncertified Discount Guess | View metric | draft · Sandbox |
Bookings vs Billings Gap | View metric | draft · Sandbox |
Illustrative pattern · synthetic data
Data-quality checks
Computed from the demo tables when the app loads them. “warn” marks mismatches this scenario includes on purpose.
| Check | Table | Rows checked | Flagged | Status |
|---|---|---|---|---|
| Invoices have a customer key | fact_invoices |
22,000 | 0 | pass |
| Invoice customer keys exist in dim_customer | fact_invoices |
22,000 | 0 | pass |
| Invoice product keys exist in dim_product | fact_invoices |
22,000 | 0 | pass |
| Invoice dates exist in dim_date | fact_invoices |
22,000 | 0 | pass |
| Invoice revenue is not negative | fact_invoices |
22,000 | 0 | pass |
| Gross margin does not exceed revenue | fact_invoices |
22,000 | 0 | pass |
| Opportunity customer keys exist in dim_customer | fact_opportunities |
4,500 | 0 | pass |
| Customer keys are unique | dim_customer |
500 | 0 | pass |
| Customers matched between CRM and ERP | dim_customer |
500 | 90 | warn |
| Every month 2023-01..2026-12 has invoices | fact_invoices |
48 | 0 | pass |
Illustrative pattern · synthetic data
Details
- fact_invoices redshift — DIST KEY(customer_key) · SORT date_key,customer_key
- fact_sales_orders redshift — DIST KEY(customer_key) · SORT order_date_key
- ext_s3_invoice_landing redshift — DIST EVEN · SORT
Illustrative pattern · synthetic data
What the AI insight is told about this platform
Platform lens: Redshift warehouse design. Emphasize dist/sort, WLM, and metric ownership.
The model also receives the synthetic KPIs, the current invoice slice and the semantic model summary.
Executive KPIs
Leadership snapshot — revenue, margin, backlog, pipeline, and samples from one reconciled model.
Revenue YTD
Gross margin %
Open backlog
On-time (open lines)
Pipeline value
Sample conversion %
Revenue & margin % by month Combo
Revenue by channel Donut
Revenue by region & channel Stacked
Top customers
Backlog by aging
What the numbers are saying
Synthetic dataset for the Amazon Redshift platform demo. Revenue and margin trends compare recent periods in this scenario. The point is leadership-ready, reconciled metrics—not a client report or live cloud tenant.
Uses the OpenAI API on synthetic demo numbers only. Public hosts apply rate limits to protect billing—private walkthroughs available on request.
Sales & margin
Where volume grows, where mix shifts, and which reps carry the book—with margin in view.
Revenue by month — top channels Stacked area
Revenue by month: latest year vs prior year YoY
Revenue by category
Sales rep leaderboard
Region performance
Customer performance
Rank, mix, and concentration—who grows revenue while eroding margin?
Customers: revenue vs. margin % Scatter
Customer rank (revenue)
Revenue by customer type
Sample-to-order
Does sampling activity convert into real revenue—and how fast?
Samples shipped by month
Conversion % by collection
Backlog & fulfillment
Where open demand is building and where fulfillment pressure is rising.
Open order value by month
Avg days late by category
CRM ↔ ERP
Pipeline stages and weighted forecast—does CRM tell the same story as orders?
Pipeline by stage
Weighted pipeline by month
Trust center
Refresh status, row counts, and quality checks that keep numbers defensible.
Last refresh
Saturday, November 21, 2026 11:43:33 AM
Row counts by table
Data quality checks
Computed from the demo tables when the app loads them. “warn” marks mismatches this scenario includes on purpose.
- pass Invoices have a customer key — 0 of 22,000 flagged (
fact_invoices) - pass Invoice customer keys exist in dim_customer — 0 of 22,000 flagged (
fact_invoices) - pass Invoice product keys exist in dim_product — 0 of 22,000 flagged (
fact_invoices) - pass Invoice dates exist in dim_date — 0 of 22,000 flagged (
fact_invoices) - pass Invoice revenue is not negative — 0 of 22,000 flagged (
fact_invoices) - pass Gross margin does not exceed revenue — 0 of 22,000 flagged (
fact_invoices) - pass Opportunity customer keys exist in dim_customer — 0 of 4,500 flagged (
fact_opportunities) - pass Customer keys are unique — 0 of 500 flagged (
dim_customer) - warn Customers matched between CRM and ERP — 90 of 500 flagged (
dim_customer) - pass Every month 2023-01..2026-12 has invoices — 0 of 48 flagged (
fact_invoices)
Semantic model
Shared tables and certified measures—the foundation for consistent answers and AI-ready Q&A.
Model tables
- dim_date
- dim_region
- dim_channel
- dim_sales_rep
- dim_customer
- dim_product
- fact_invoices
- fact_sales_orders
- fact_inventory_backlog
- fact_samples
- fact_opportunities
- bridge_crm_erp_keys
- fact_data_quality_events
- fact_pipeline_runs
- dim_semantic_measure
- fact_medallion_stats
- dim_warehouse_object
- demo_scenario
- data_dictionary
Certified measures
- Total Revenue
- Revenue YTD
- Gross Margin
- Gross Margin %
- Order Value
- Open Backlog
- Pipeline Value
- Weighted Pipeline
- Samples Shipped
- Converted Samples
- Sample Conversion %
Example natural language questions
- What is revenue YTD by channel?
- Which customers grew backlog the most?
- Show margin % trend by month.
- Who are the top reps by pipeline value?
- What is the CRM↔ERP match rate?
Illustrative pattern · synthetic data
How to present this