Supply Chain Control Tower for Retail Giant: How We Unlocked €2M+ in Hidden Inventory

Executive Summary

A top European retailer (400+ stores, 20M online customers) was losing millions due to blind spots in its supply chain: inconsistent numbers, conflicting KPIs, and logistical bottlenecks no one could pinpoint.

We built a Supply Chain Control Tower - a single operational truth for more than 100 planners. For the first time, leadership could see the full lifecycle of every product: supplier → warehouse → store → customer.

Key Outcomes:

• €2M+ in unnecessary inventory identified and corrected

• Stockout root‑cause clarity → fewer avoidable shortages

• Decision‑making cycle time reduced from weeks  →  days

• 100+ supply chain users operating on a single source of truth

• 80% reduction in Analysis Services load → faster, reliable reporting

• Onboarding time cut from weeks → days

Client Profile and Context

The client operates 400+ physical stores and a significant online presence with 20 million customers, employing 4,000 people with 2,000+ using data platforms. The supply chain department consisted of 100 team members organized into specialized sub-teams (demand planning, stock optimization, supplier management), with 40 active Power BI users possessing advanced SQL skills and strong self-service capabilities.

Business Challenges

Decentralized Data Architecture and Logic

The supply chain team operated as a fully independent data island. Each sub‑team extracted its own ERP data, built its own logic, and maintained its own reports. The result: different numbers for the same KPI, no shared definitions, and no reliable view of where products were actually stuck.

No Single Source of Truth

Inventory levels, order statuses, and warehouse positions looked different depending on who was asked and which system they used. Critical knowledge lived in individual spreadsheets and in the heads of a few experts. Leadership couldn’t rely on any single version of the truth.

Complex Source System Environment

Data came from everywhere - ERP, an aging SQL server, local databases, bulk exports, Excel files - none of them aligned. The central warehouse platform existed, but no one trusted it. Reports broke easily, and uncovering operational bottlenecks was nearly impossible.

Inefficient Analysis Services Usage

The Azure Analysis Services model was slow, frequently outdated, and missing critical supply chain definitions. Users began downloading massive data volumes for local processing, a usage pattern that degraded performance for everyone while still not meeting their needs.

Limited Accessibility

Only 2-3 specialists per plant could access critical systems due to licensing constraints and technical complexity. When these individuals were unavailable, essential reporting simply stopped.

Solution Design

Astral Forest built a Supply Chain Control Tower consolidating all operational data into one consistent, governed, business‑owned source of truth.

Key principles:

• Business‑owned definitions: one KPI, one meaning

• Full lifecycle visibility: from supplier to warehouse to shelf

• Cross‑team transparency: every decision maker sees the same reality

• Self‑Service at scale: 100 people working from one shared model

To ensure a seamless transition, we introduced a Playground environment allowing teams to migrate without downtime or operational risk. This preserved continuity while building trust in the new model.

The results: The Control Tower revealed what leadership suspected but couldn’t prove: Most stockouts were caused internally, not by suppliers. For the first time, the team could distinguish:

• items stuck in transit

• items stuck in internal warehouses

• items blocked by commerce rules

• items missing due to operational errors

This insight alone unlocked over €2M in hidden inventory leakage and drove targeted process improvements across the entire supply chain network.

Implementation Approach

The transformation worked because the supply chain - not IT - owned the truth.

Business Domain Ownership

Each part of the supply chain (supplier orders, inventory accuracy, store replenishment) received clear end‑to‑end ownership.This ensured every domain delivered real operational value before the team moved to the next one.

Cross‑Functional Delivery

Instead of separating business and IT, we created one joint delivery group:

• Business leaders made decisions on KPI definitions and logic.

• Analysts and planners validated the data and the insights.

• Our team ensured the new platform reflected real operational workflows.

This structure turned data quality from “someone else’s problem” into a leadership responsibility.

Full Transparency & No Disruption

All existing reporting continued to run throughout the transformation. There was no big‑bang switch, no downtime, and no risk to daily operations. Teams could compare old vs. new until they fully trusted the unified source of truth.

Key Challenges and Solutions

1. Retiring a Critical Legacy System Without Disrupting Operations: The legacy database was deeply embedded in daily processes. Removing it without breaking the business required a controlled, audited transition. We matched every use case with an improved alternative, validated each with the business, and only then retired the system.

Outcome: zero disruption, zero firefighting.

2. Fragmented Knowledge and Conflicting Definitions: Much of the company’s operational knowledge lived inside spreadsheets or in the heads of a handful of experts. Through structured sessions with supply chain teams and the ERP provider, this knowledge became formal, governed business logic.

Outcome: zero disruption, zero firefighting.

3. No Clear View of Where Products Were Stuck: Data from suppliers, warehouses, stores, and ERP systems didn’t align. By integrating all operational signals into a single lifecycle view, leadership could finally answer: “Is the issue at the supplier, in transit, in the warehouse, or in store operations?”This enabled targeted action instead of assumptions.

Outcome: stockout causes became measurable and solvable.

4. Five Sub‑Teams, Five Versions of Reality: Different groups used different KPI definitions, often contradicting one another. We facilitated alignment where possible and created transparent, documented variants where legitimate differences existed.

Outcome: collaboration replaced finger‑pointing.

5. Resistance to Change: Operational teams feared losing control and visibility. We demonstrated improvements using real business scenarios and allowed parallel usage of both systems until trust was built.

Outcome: fast adoption by 40+ advanced users.

Business Impact and Results

Operational Improvements

• Report creation and decision cycles dropped from weeks to days.

• New team members ramped up in days, not weeks, thanks to clarity and consistency.

• 100+ people now operate using the same single source of truth.

• All power users moved to the new platform within six months.

Advanced Analytics: The Presence Rate Example

The Presence Rate report became the executive dashboard that changed how leadership understood stockouts. It measured one critical question:

"Is the product physically available in the warehouse for the online order?"

This sounds simple. It's not. This single metric required stitching together data from ERP, WMS, e-commerce platforms, supplier delivery systems, and warehouse operations—data that had never been connected before.

The report categorized why products were out of stock across multiple dimensions:

• Product categories: Best-sellers, core products, novelties, secondary catalog

• Time-based trending: Identifying seasonal patterns and concerning degradations

• Store-level analysis: Prioritizing top 20 stores by customer volume

• Supplier performance: Revealing which suppliers consistently delivered expected inventory

Most powerfully, the report categorized why products were out of stock:

• Commerce-related: Products unavailable for ordering or imports not yet in country

• Supplier-related: Items in transit (on schedule, delayed, or partial delivery)

• Supply chain internal: Products in warehouse but not delivered to stores

• Warehouse operational: Process issues at distribution centers

• Uncategorized: Cases requiring specific investigation

This revealed that most stockout problems were internal rather than supplier-caused - uncomfortable but actionable insight only possible because the team understood data models deeply enough to trace complet order lifecycles.

User Feedback

Supply chain team members consistently noted:

• Writing SQL queries against star schema was easier than DAX formulas

• Lower entry barrier for new members accelerated onboarding

• Consistent naming across tables eliminated confusion

• Detailed documentation provided self-service answers

• Analyzing large data volumes was significantly easier

• Users recognized star schema benefits despite initial resistance

Lessons Learned

What Worked Well

1. End-to-end domain ownership: Product owners delivered complete, coherent domains instead of fragmented features

2. Formal business decision makers: Accountability transformed from "IT's problem" to "business leadership responsibility"

3. The Playground Schema: Enabled gradual migration without disruption - users migrated when ready, not when forced

4. Documentation as a first-class deliverable: Heavy investment in documentation paid dividends in adoption and self-service

5. Leveraging existing SQL skills: Rather than forcing new tools, we made their existing skills more powerful

6. ELT architecture: Centralized logic in one version-controlled location, offloaded source systems

‍

Recommendations for Similar Projects

• Start comprehensive documentation from day one. It becomes critical for adoption. Don't treat it as "we'll document later"- you won't.

• Invest significant time understanding source system logic deeply. Surface-level knowledge creates future problems when edge cases emerge.

• Identify and formally empower business decision makers early with clear accountability. Without this, nobody owns data quality.

• Consider transitional approaches like Playground Schema for large migrations to reduce risk. "Big bang" cutovers rarely work.

• Don't underestimate cultural change required. The technical solution is only 30% of the challenge. The other 70% is people.

• Plan for education on new data structures, especially star schema. Users won't adopt what they don't understand.

• Build performance optimization into development, not after go-live. Slow systems kill adoption faster than missing features.

• Create feedback loops with end users throughout development. Don't wait until "final delivery" to discover you built the wrong thing.

• Be prepared for initial resistance and have concrete benefit demonstrations ready. Show, don't tell.

Key Metrics Summary

Data Structure:

• 30 dimension tables, 20 fact tables, 5 bridge tables

• 50+ Playground Schema objects (initially, decreasing over time)

• 20+ terabytes total data storage

Performance Improvements:

• 80% reduction in Analysis Services load

• Report development: weeks → days

• Query response: 30-60 seconds → 3-5 seconds

• Daily refresh completion before 8 AM guaranteed

Adoption:

• 100+ supply chain members with access

• 40 active Power BI developers

• 200+ reports migrated within 6 months

• 90%+ user satisfaction post-implementation

Business Impact:

• Single source of truth across all supply chain processes

• Stockout root cause attribution enabling targeted improvements

• Inventory optimization reducing costs while improving availability

• New member onboarding: weeks → days

Conclusion

The Supply Chain Control Tower did more than unify data - it restored operational control across a complex retail network. For the first time, leadership could see exactly where products were getting stuck, why availability dropped, and how internal processes contributed to losses.

This clarity unlocked over €2M in hidden inventory leakage, reduced avoidable stockouts, and accelerated decision-making from weeks to days. But the most significant impact happened inside the organization. Teams stopped debating numbers and started solving problems. Planners, analysts, and managers aligned around one shared version of reality. Accountability became data-driven. Trust in reporting - lost over years - was rebuilt through transparency and consistent logic.

The Control Tower is now the operational backbone for the retailer’s supply chain. It supports daily decisions, powers strategic reviews, and provides the foundation for future AI-driven forecasting, optimization, and automation.

This transformation was not a technology upgrade. It was a structural shift that gave leadership full visibility, governance, and control over product availability - and the financial impact that comes with it.