Mapping the retail data landscape end to end

2. Warehouse and logistics data. Captured by the WMS (Warehouse Management System), covering putaway, pick paths, inventory location, and cycle counts. Paired with TMS (Transportation Management System) data for shipping and carrier performance.

3. Point-of-sale (POS) and transaction data. The classic retail dataset: SKU, price, quantity, timestamp, store ID, tender type. This is the most mature data source in most retailers, often 20 to 30 years deep in legacy systems.

4. Customer and loyalty data. Loyalty program IDs, CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète → (Customer Relationship ManagementCustomer Relationship ManagementCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète →) records, email and app engagement, and increasingly, unified customer profiles built from a CDPCDPA Customer Data Platform unifies customer data from all sources into persistent, actionable profiles that other systems can use.Voir la définition complète → (Customer Data PlatformCustomer Data PlatformA Customer Data Platform unifies customer data from all sources into persistent, actionable profiles that other systems can use.Voir la définition complète →, software that consolidates customer data from multiple sources into a single profile).

5. Digital and clickstream data. Website and app sessions, product views, cart adds, search queries, and ad exposure. Captured via tools like Google Analytics or Adobe Analytics, and typically the least integrated with in-store data.

A sixth, cross-cutting category: inventory and merchandising master data, the product catalog, attributes, pricing, and promotions that should tie every other dataset together, but frequently doesn't, because SKUs are coded differently across systems.

Why these sources rarely connect

Most retailers run on a patchwork built over 15 to 20 years: a legacy POS platform, a separately-sourced WMS, a bolted-on e-commerce stack, and a CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète → acquired through M&A. Each has its own product ID scheme, customer ID scheme, and update cadence. The result is what practitioners call "identity fragmentation": the same customer, product, or store exists as different keys in different systems, and nobody owns the mapping.

This is why "omnichannelomnichannelAn integrated approach connecting all customer touchpoints (physical, digital, mobile) into a seamless experience, with shared data and consistent context across channels.Voir la définition complète →" analytics is hard in practice, not because the concept is unclear, but because the underlying joins don't exist without heavy investment in master data managementmaster data managementMaster Data Management (MDM) is the discipline of creating and maintaining a single, consistent, trusted version of an organization's core business entities like customers, products, and suppliers.Voir la définition complète → (MDMMDMMaster Data Management (MDM) is the discipline of creating and maintaining a single, consistent, trusted version of an organization's core business entities like customers, products, and suppliers.Voir la définition complète →), the discipline and tooling for maintaining a single trusted version of core entities like product, customer, and location.

Data qualityData qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.Voir la définition complète →: the metrics that actually matter

Data qualityData qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.Voir la définition complète → isn't binary. It's measured across dimensions, and retail has sector-specific failure points for each.

• Completeness: Are required fields populated? A common retail failure: 15 to 20% of SKUs missing key attributes (size, color, material) at launch, estimate based on typical PIM (Product Information Management) audits.
• Accuracy: Does the data reflect reality? Inventory accuracy (system count vs. physical count) is a classic benchmark. Best-in-class retailers hold this above 95%; many run closer to 85-90%, according to industry surveys from groups like RILA (Retail Industry Leaders Association).
• Timeliness: How stale is the data when used? Real-time inventory visibility (updated within minutes) versus batch updates (once daily) materially changes whether "in stock" online means in stock in reality.
• Consistency: Does the same entity match across systems? A product with three different SKU codes across ERP, WMS, and e-commerce is a consistency failure.
• Uniqueness: Are there duplicate records? Duplicate customer profiles are endemic in loyalty databases, often 10 to 15% of records, estimate, before deduplication.

A worked example: the cost of bad inventory data

Say a mid-size apparel chain has 90% inventory accuracy across 10,000 SKUs. That means roughly 1,000 SKUs show an incorrect on-hand count at any time.

If even a quarter of those errors cause an online "in stock" promise that can't be fulfilled (a "phantom stock" scenario), that's 250 SKUs generating failed promises. If each failed promise costs one lost sale averaging $40, that's a rough $10,000 exposure per inventory snapshot, compounding daily until reconciled. This is a simplified illustrative calculation, not a benchmark, but it shows why inventory accuracy is treated as a governance-level KPIKPIKey Performance Indicator, a measurable value that shows how effectively you're achieving a specific objective, tracked over time against a target.Voir la définition complète → (Key Performance IndicatorKey Performance IndicatorKey Performance Indicator, a measurable value that shows how effectively you're achieving a specific objective, tracked over time against a target.Voir la définition complète →), not just an operations metric.

Governance: who owns the fixes

Data governanceData governanceData governance is the set of policies, roles, and processes that ensure data is accurate, secure, well-defined, and used responsibly across an organization.Voir la définition complète → in retail typically centers on:

• Data stewardshipData stewardshipA business-side owner responsible for the quality, consistency and appropriate use of data in their domain.Voir la définition complète →: named owners per domain (product, customer, location) accountable for quality thresholds.
• Master data management (MDM): the technical and process layer resolving duplicate or conflicting records into a "golden record."
• Privacy and compliance: in the EU, the GDPR (General Data Protection Regulation) governs customer data collection and consent; in the US, state laws like the CCPA (California Consumer Privacy Act) and its expansion, the CPRA, set similar rules. Retailers handling loyalty and clickstream data must document lawful basis for processing and honor deletion requests, a governance obligation, not just a legal one.

Analytics and measurement benchmarks

Once data is inventoried and reasonably trustworthy, the next question is: what should you actually measure, and against what benchmark?

Core retail analytics KPIs

• Sell-through rate: units sold ÷ units received, over a period. A healthy apparel sell-through in the first 4 to 6 weeks of a season is often cited around 70-80%, estimate, varying heavily by category.
• Inventory turnover: cost of goods sold ÷ average inventory value. Grocery retailers often run turnover of 10 to 15x annually; apparel and durable goods run much lower, often 3 to 5x, estimates that vary widely by subsector.
• Conversion rateConversion rateThe percentage of visitors or prospects who complete a desired action (purchase, sign-up, contact form), calculated as conversions divided by total opportunities.Voir la définition complète →: transactions ÷ visits (in-store, via people-counters, or online). E-commerce conversion in the US is commonly cited around 2 to 3%, estimate as of recent years, per benchmarks from firms like Adobe Digital Economy Index.
• Customer match rate: the percentage of transactions successfully linked to a known customer identity. This underpins any loyalty or personalization analytics and is often below 50% in-store, versus near 100% for logged-in e-commerce sessions.

A simple data-quality-to-KPIKPIKey Performance Indicator, a measurable value that shows how effectively you're achieving a specific objective, tracked over time against a target.Voir la définition complète → snippet

Analysts often start quality diagnostics with basic completeness and duplicate checks before trusting a KPIKPIKey Performance Indicator, a measurable value that shows how effectively you're achieving a specific objective, tracked over time against a target.Voir la définition complète → dashboard:

-- Completeness check: % of SKUs missing a required attribute
SELECT
  COUNT(*) AS total_skus,
  SUM(CASE WHEN size IS NULL OR color IS NULL THEN 1 ELSE 0 END) AS incomplete_skus,
  ROUND(100.0 * SUM(CASE WHEN size IS NULL OR color IS NULL THEN 1 ELSE 0 END) / COUNT(*), 1) AS pct_incomplete
FROM product_master;

Running this before building a sell-through dashboard prevents a common failure mode: reporting a KPIKPIKey Performance Indicator, a measurable value that shows how effectively you're achieving a specific objective, tracked over time against a target.Voir la définition complète → confidently on top of a broken foundation.

Benchmarking mindset

Benchmarks are directional, not universal targets. A grocery chain's inventory turnover target is meaningless applied to a luxury fashion house. Always benchmark within subsector, and treat every externally-cited figure as an estimate unless it comes from your own audited systems.

🎬 [VIDEO: "How Retailers Use Data Analytics" - youtube.com/@McKinsey - a concise overview of how leading retailers connect data sources to decision-making, useful for grounding the concepts in this lesson]

Key Takeaways

• Retail data comes from at least five distinct source clusters (supply chain/EDI, warehouse/WMS, POS, customer/CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète →, digital/clickstream), and most retailers have never fully mapped or joined them due to identity fragmentation across legacy systems.
• Data qualityData qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.Voir la définition complète → should be assessed across five dimensions: completeness, accuracy, timeliness, consistency, uniqueness. Inventory accuracy below 90% is common and has direct, calculable revenue impact.
• Governance requires named data stewards, MDM to resolve duplicate "golden records," and compliance with GDPR (EU) and CCPA/CPRA (US) for customer data.