Luxury Beauty · Data Ownership Case Study

Sales & Distribution
Data Control Tower

Reliable sell-out, governed point-of-sale master data, 1-3 year network planning and executive performance analytics across a synthetic multi-market beauty landscape.

A reproducible operating model connecting SAP-shaped master data, Data Entry v2, retailer feeds, SimpliField audits, GCP/BigQuery, Anaplan planning, Newviz reconciliation, Excel and Power BI.

240 POS12 marketsSell-out reliabilitySCD Type 2Data Entry v2GCP / BigQueryAnaplanPower BIExcel + PythonQA & UAT

Synthetic data only. Independent methodological case study; no affiliation with or access to any company system.

Control Tower SnapshotSynthetic model · v1.0
Sell-out analysed
€112.50m
Latest coverage
97.81%
Curated critical DQ
100.00%
Forecast WAPE
9.21%
Actual vs approved plan
Actual through Jun 2026 Approved plan Jul 2026 → Dec 2029
Executive KPIs

Control tower at a glance

Ten headline metrics summarising scope, commercial scale, data reliability, planning quality and operating support.

Markets
12
Points of sale
240
Sell-out analysed
€112.50m
Curated sell-out rows
45,708
Coverage - latest
97.81%
Eligible active POS
On-time - latest
97.68%
Critical DQ - curated
100.00%
Forecast WAPE
9.21%
Time-based holdout
Support SLA met
93.89%
Reconciliation after replay
0.00%

Coverage uses eligible active POS. WAPE uses the March–June 2026 time-based holdout.

End-to-end operating model

From source event to trusted management decision

P0Step 1
Scope and governance
Metric dictionary, RACI and publish gates
P1Step 2
Synthetic source landscape
SAP, Data Entry v2, retailer, SimpliField and Newviz extracts
P2Step 3
Distribution master
Stable keys, GLN-shaped identifiers and SCD2 history
P3Step 4
Sell-out quality
Eight rules, correction log, quarantine and audit trail
P4Step 5
GCP transformation
Raw-curated-mart SQL and Dataform assertions
P5Step 6
Cross-system reconciliation
SAP-Anaplan and GCP-Newviz controls with replay
P6Step 7
Forecast and planning
9.21% WAPE baseline and approved 1-3 year plan
P7Step 8
Executive analytics
Excel model, DAX, Power Query and report specification
P8Step 9
Support model
Tickets, priority SLA and operating procedures
P9Step 10
QA and release
Automated tests, UAT evidence and SHA-256 manifest
Reference architecture
SAP-shaped POS master → GCP raw
Data Entry v2 / retailer feeds → GCP raw
SimpliField audits → GCP raw
GCP raw → GCP curated: approved corrections, DQ, SCD2
GCP curated → GCP marts
GCP marts → Newviz / Power BI / Excel
GCP marts ↔ Python + Anaplan forecast and plan

Raw evidence is preserved. Critical curated controls must pass before any consumer refresh.

Sell-out reliability

Coverage, timeliness and quality are managed as one service

Sell-out coverage and on-time submission
Monthly, 18 months. 95% target dotted.
Critical DQ rules: raw vs curated
Five critical rules, 96%-100% axis.
Coverage
80.34% → 97.81%
+17.47 pp
Record-level critical quality
98.11% → 100.00%
After controlled remediation
Timeliness
97.68%
Latest month, above 95% target

The Excel workbook also shows a 99.62% weighted rule-level first-pass rate; record-level and rule-level rates answer different questions.

A late consumption batch is detected, replayed and closed

Before replay
0.1761%
GCP–Newviz delta · status INVESTIGATE
Threshold 0.10% absolute monthly delta.
After replay
≈ 0.0000%
Residual delta · status PASS
POS-month full outer join → affected keys → replay → rerun control → publish.
Distribution master

One point-of-sale identity across SAP, Anaplan and GCP

240
Current POS records
262
SCD2 versions
228
Active POS at Jun 2026
97.08%
Initial SAP–Anaplan match
10
Controlled mismatches queued
SCD Type 2 example
Previous version
valid_to = 2026-03-31
current_flag = false
Current version
valid_from = 2026-04-01
current_flag = true

Stable natural key · surrogate version key · valid_from · valid_to · current_flag.

POS creation / change workflow
  1. 1Market request
  2. 2Data Owner uniqueness / reference checks
  3. 3SAP Business Partner / customer creation or change
  4. 4GCP SCD2 merge
  5. 5Anaplan and Data Entry activation
  6. 6Four-system reconciliation and publish gate
System roles
System
Role
Evidence
SAP
Source of truth
POS master creation SOP
GCP
SCD2 history
Curated dim_pos with versions
Anaplan
Planning reference
Import module mapped to SAP key
Data Entry
Eligible POS list
Feed contract + JSON schema
Newviz
Consumer
Cube reconciled to GCP marts
Commercial performance

€112.50m of synthetic sell-out across 12 markets

Synthetic values only — figures illustrate the operating model, not real company performance.

Sell-out by market
Actual monthly sell-out
Through Jun 2026, actuals only.
€6.49m
Latest month
75.06%
Synthetic gross margin
1.29 mo
Weighted stock cover
Forecast & 1-3 year plan

Statistical baseline, network plan and governed management challenge

9.21%
Holdout WAPE
+2.34%
Bias
€16,784
MAE / market-cat-month
30 / 15
Openings / closures to 2029
ScenarioClient-side recalculation from approved_plan_eur. Source JSON is not mutated.
2026
€50.19m
Current forecast · H2 only
2027
€101.05m
Budget · 1Y
2028
€113.34m
Strategic plan · 3Y
2029
€128.89m
Strategic plan · 3Y
Approved plan formula
Approved plan = statistical baseline
              + network uplift
              + baseline × (market override + category mix override + central challenge)
Grain
Market · category · month
Features
Seasonality and lag features
Model
Gradient Boosting
Holdout
March–June 2026
Blend
35% model signal + 65% prior-year anchor
Implementation evidence

Every tool is represented by a concrete implementation asset

SAP S/4HANA
implementation-ready pack
POS source-of-truth mapping and creation/change SOP
Anaplan
implementation-ready pack
DISCO-style blueprint, imports, workflow, 1-3 year plan
GCP / BigQuery
offline parity executed
partitioned SQL, SCD2, marts, reconciliation
Dataform
source package
incremental models and assertions
Data Entry v2
contract and simulator
JSON schema, error catalogue and UAT
Newviz
consumption contract
cube contract and GCP reconciliation
SimpliField
synthetic extract
retail audit form and 2,326 observations
Python
executed
generation, controls, forecast, assets and tests
Excel
executed and rendered
formula-driven FP&A and data-control workbook
Power BI
implementation-ready pack
Power Query, 40+ DAX measures, theme and report design
Excel
18 sheets · 14,964 live formula cells · MODEL STATUS PASS
Power BI
46 explicit DAX measures · Power Query schema contract · 5-page report design
GCP / Dataform
10 SQL / SQLX / workflow assets · incremental model · assertions
Python
Deterministic generation · DQ · forecast · assets · automated tests

The data product includes an operating service, not only code

180
Synthetic tickets
93.89%
SLA met
9.43h
Median resolution
6 categories
Late feed · POS mismatch · Data Entry · Access · Newviz recon · Anaplan import
Daily
Monitoring
Weekly
Mapping / coverage review
Monthly
Close · reconciliation · forecast
Budget cycle
Plan approval
Quarterly
Control review
Methodology

Concrete rules, formulas and public references

Net sell-out EUR
(gross local − discount local − returns local) / local-per-EUR FX
Coverage
distinct reporting POS / eligible active POS
Forecast WAPE
sum |forecast − actual| / sum actual
Bias
sum(forecast − actual) / sum actual
GCP–Newviz delta
(GCP EUR − Newviz EUR) / GCP EUR
Ending active POS
beginning POS + openings − closures
QA & reproducibility

Release evidence is explicit and downloadable

12 / 12 automated Python tests PASS
30 / 30 UAT cases PASS
240 / 240 SCD2 controls PASS
9 / 9 Excel release checks PASS
0 failed critical curated rules
No Excel formula error token
Post-replay reconciliation PASS
Deterministic seed 20260712
Repository

The full model, code and evidence are public.

Synthetic data only. Independent methodological case study; no affiliation with or access to any company system.

Open the repository