Skip to main content

FMCG & Packaging Analytics

Trade Spend at 20% of Revenue. ROI Visibility at Zero.

FMCG companies run seven to twelve disconnected systems - ERP, DMS, TPM, WMS, CRM - none of which agree on the same SKU list. We connect them, clean them, and give your supply chain, commercial, and finance teams the data they need to make decisions during the period, not after it closes.

20–40%Forecast accuracy gain
8%Industry stockout rate
7–12Disconnected systems
Book an FMCG Data AssessmentFMCG & Packaging Overview

8%

Industry average stockout rate - despite "adequate" inventory on paper

15–25%

Of net revenue absorbed by trade spend - with no ROI measurement

7–12

Disconnected systems per FMCG company - ERP, DMS, TPM, WMS, CRM - none agree

The Real Problem

The data exists. The problem is it's in seven systems that have never spoken to each other.

Most FMCG companies we work with are not short of data. Sell-in is in the ERP. Sell-out is in the distributor portal. Trade promotion plans are in the TPM system. Retailer deductions are in the finance inbox. Packaging line performance is in the MES - or in a shift supervisor's spreadsheet that someone built in 2018 and everyone uses but nobody owns.

None of it talks to anything else. So the demand forecast is driven by sales targets rather than market signals. The S&OP meeting is a data argument rather than a decision forum. Trade promotions are planned without knowing what the last promotion returned. And the stockout that frustrates your retail buyer happens alongside excess stock of slow-moving variants that nobody had the visibility to redirect.

We fix the data foundation first. One SKU master. One demand signal. One commercial truth. When your supply chain, trade marketing, and finance teams are reading from the same source, the decisions they make - and the speed at which they make them - change materially.

Seven to twelve systems, zero shared truth

ERP, DMS, TPM, WMS, CRM - each holding a fragment. None of them connected. Reconciling them is a weekly exercise in data archaeology.

S&OP meetings debating numbers, not making decisions

Sales have one number. Supply chain have another. Finance have a third. The meeting exists to make decisions but spends its time arbitrating data disputes.

Trade spend planned blind - no sell-through data from the last promotion

Retailer sell-out data arrives weeks late, in formats that don't match the ERP. Each promotion is planned without knowing what the previous one returned.

Stockouts alongside excess - both caused by the same broken forecast

Fast movers are out at shelf. Slow-moving variants are clogging the warehouse. One forecast. Wrong in both directions simultaneously.

Who Feels the Pain

Every function in your FMCG business has a data problem. Each one is different.

Select your role to see the specific challenges your team faces - and exactly what MDI builds to resolve them.

Supply Chain Director

“Adequate inventory. Chronic stockouts. Both at the same time.”

What We Hear Every Time

1

Demand forecasts driven by sales targets, not statistical models

The S&OP process starts with what sales want to sell, not what the market is actually signalling. Bias gets baked in at the source - so the forecast is always optimistic on the way in, always wrong on the way out. Nobody's model is using actual sell-through data. Everyone's arguing about whose number is right rather than fixing the number itself.

2

Eight per cent stockout rate despite "adequate" inventory on paper

The aggregate inventory position looks fine. The problem is that the wrong SKUs are holding the stock. Fast movers at the shelf are out. Slow-moving variants are clogging the warehouse. Without SKU-level days-of-cover visibility that's live and channel-specific, replenishment decisions are always slightly wrong - and slightly wrong, repeated daily across 300 SKUs, becomes structurally expensive.

3

Seven to twelve systems with different SKU lists and no shared truth

ERP, DMS, TPM, WMS, CRM, retailer portals - none of them agree on the same SKU hierarchy, the same customer code, or the same inventory figure. Reconciling them is a weekly exercise in data archaeology. The supply chain team spends half its analytical capacity just getting the data to agree before they can use it.

What We Build For You

We unify sell-in and sell-out data across your ERP, distributor management system, and retailer feeds into a single demand intelligence layer. Statistical forecasting models replace the sales-biased baseline - removing the optimism and the arguments simultaneously. Days of cover is tracked live by SKU and by channel. Replenishment signals trigger automatically, before the stockout, not because of it. The S&OP meeting stops being a debate and starts being a decision.

›

Demand forecast accuracy improved 20–40% when sales bias is removed from the baseline

›

Days of cover live by SKU and channel - replenishment before the stockout

›

Single SKU master across ERP, DMS, and retailer feeds - one source of truth

Book a Supply Chain Director Assessment

The MDI Operating Model

Unify → Predict → Act. Applied to FMCG.

We don't start with a dashboard. We start with the data foundation. Then we layer in prediction. Then we automate the action. In that order - every time.

01

UNIFY

One SKU Master Across Every System

We connect your ERP, distributor management system, trade promotion management platform, WMS, and retailer data feeds into a single governed data layer on Microsoft Fabric. The seven to twelve systems that have never agreed on a SKU list, a customer code, or an inventory figure are reconciled into one version of truth - with lineage, schema enforcement, and a master data layer that all functions read from.

  • ›ERP + DMS + TPM integration
  • ›Retailer sell-out feeds and portal extracts
  • ›Unified SKU master across all source systems
  • ›SAP, Oracle, NetSuite, Dynamics & other ERP connectors

02

PREDICT

Forecasts Built on Signals, Not Sales Targets

On a unified data foundation, we deploy statistical demand models that read from actual sell-out signals - not from what the sales team is targeting. Forecast accuracy improves 20–40% when the bias is removed from the baseline. Promotion lift is modelled from historical sell-through, not estimated after the fact. Days of cover is forecast by SKU and channel, so replenishment calls are made before the stockout, not because of it.

  • ›Statistical demand forecasting - MAPE and Bias tracked by SKU
  • ›Promotion baseline and lift modelling from sell-out data
  • ›Days-of-cover forecast by SKU and channel
  • ›Slow-mover and excess inventory flagging

03

ACT

Replenishment Triggered, Not Requested

Insight without a closed loop is just reporting. We connect the demand signal to the replenishment action - automated alerts to procurement, push notifications to the commercial team when sell-through falls behind promotion targets, and executive dashboards that reflect commercial reality in real time, not after the period closes.

  • ›Automated replenishment alerts to procurement
  • ›Sell-through deviation alerts during promotions
  • ›S&OP data layer - one agreed number across functions
  • ›Executive commercial dashboard updated daily

Capabilities

Six analytics capabilities that move the metrics that matter in FMCG.

20–40%

forecast accuracy gain

Demand Forecasting & S&OP

Statistical forecast models built on sell-out data, not sales targets. MAPE and Bias tracked by SKU, by channel, and by period. The S&OP meeting gets one agreed number - not three versions from three functions that spend the first hour arguing about which is right.

15–25%

of revenue in trade spend

Trade Promotion ROI Analytics

Baseline calculated from clean historical sell-out. Lift measured during the promotion - not estimated afterwards. Sell-through tracked by customer and SKU. The next planning cycle is informed by what the last one actually returned, not by what it was supposed to return.

60 sec

update frequency

Packaging Line OEE

OEE at machine level on packaging lines - updated every 60 seconds, not compiled at shift end. Changeover sequences captured event by event for genuine benchmarking across crews. Yield and wastage attributed to the specific run, SKU, and material batch that generated it.

Monthly

not quarterly

SKU Profitability Intelligence

Cost per case by SKU - factoring in production cost, changeover time, packaging material variance, trade spend, and freight. Margin by channel, by customer, and by period. Net revenue management moves from a strategy document to an operating discipline.

Live

during the promotion

Sell-Through & Customer Analytics

Sell-through rate by customer account, by SKU, and by week. Execution quality tracked during the promotion - before the deductions arrive. The accounts that execute well separated from those that do not, with data that informs the next trading terms conversation.

Before

the stockout

Automated Replenishment Workflows

Replenishment signals triggered from live days-of-cover calculations - not from monthly average consumption. Purchase order drafts raised automatically in the ERP when stock falls below the dynamically calculated reorder point. Power Automate workflows connecting the signal to the action.

Proof Stats

Numbers that come from FMCG operations, not analyst reports.

20–40%

Demand Forecast Accuracy Improvement

When sales bias is removed from the baseline and the model is built on actual sell-out signals, not sell-in targets.

15–25%

Of Net Revenue in Trade Spend

The industry norm. ROI visibility on that spend is not. Companies that measure promotion lift outperform those that don't by 3–5% in net revenue.

7–12

Disconnected Systems per FMCG Company

ERP, DMS, TPM, WMS, CRM, retailer portals - none of which agree on the same SKU list. Reconciling them manually is the hidden cost of every planning cycle.

Hollandia Dairy

Dairy / FMCG · USA

Case Study

Challenge

Demand data fragmented across trade channels - no consolidated view of sell-out versus sell-in by SKU. Spoilage and stockouts managed reactively. Replenishment based on gut feel and phone calls.

Solution

Unified ERP, distributor channel feeds, and cold chain logistics into a single demand intelligence layer. Statistical forecasting on actual sell-out signals. Automated replenishment alerts. Spoilage made predictable before it became a write-off.

Results

5–15%

Sales Uplift

20–40%

Stockout Reduction

10–30%

Spoilage Reduction

Kim Moten, Programme Director

Verified · Clutch.co · 2024

Full case study

Technology Stack

Built on Microsoft Fabric. Integrated with your existing systems.

Microsoft Fabric

Unified data platform - Lakehouse, Delta Lake, OneLake

Power BI

Commercial dashboards - S&OP, trade, SKU profitability

Azure Data Factory

ERP, DMS, TPM, and retailer data integration

ERP Integration (SAP / Oracle / NetSuite / Dynamics)

Read-only connectors - no ERP changes required

Power Automate

Replenishment workflows and promotion alerts

Copilot Studio

Commercial analytics assistant for the S&OP team

Delta Lake

Versioned, governed SKU and transaction data

Azure ML

Statistical demand forecasting and promotion lift models

Common Questions

Questions FMCG teams ask before they engage us.

How long does it take to get a demand forecast model live for our SKU range?

Typically 8–10 weeks from discovery to a live statistical forecast model. Week 1–2: data audit across ERP, DMS, and retailer feeds. Week 3–4: SKU master data cleansing and unified data layer. Week 5–7: model build and validation against historical actuals. Week 8–10: Power BI deployment, S&OP integration, and team training. The first S&OP cycle using the model is the real test - we are present for it.

Can you connect to our ERP and distributor management system without changes to either system?

Yes. We connect using read-only extraction - no configuration changes to your ERP or DMS. This applies across SAP, Oracle, NetSuite, Microsoft Dynamics, Sage, and other platforms. Data is pulled on a scheduled or near-real-time basis depending on the use case. The analytics layer is fully separate from your transactional systems.

What sell-out data sources can you integrate?

Distributor management systems (DMS), retailer portal extracts (Carrefour, Lulu, Spinneys, and others), EDI feeds from key accounts, and syndicated data sources where available. We normalise all of these against a common SKU and customer master - so every sell-out signal feeds the same demand model regardless of source format.

Our trade promotion data is in a legacy TPM system. Can you integrate it?

Yes. We have integrated with major TPM platforms as well as Excel-based promotion trackers. The promotional calendar, planned spend, and target volumes are connected to actual sell-through data - so the ROI calculation runs from the same source as the commercial result, not from two separate systems that disagree on the baseline.

What does Phase 1 look like for an FMCG company that has never done this before?

Phase 1 is typically a 90-day sprint. We recommend starting with the demand foundation: connect 2–3 source systems, build a clean SKU master, and deploy a statistical forecast model for your top 50 SKUs. This gives you a proof point in the next S&OP cycle before you commit to a full programme across the entire range and all channels.

Start Here

Book a 30-minute FMCG Data Assessment.

We'll review your current system landscape, identify where the forecast bias is coming from, and tell you what a Phase 1 looks like for your specific environment. No slides. No generic deck.

Book the AssessmentSee All Case Studies
✓

30 minutes - your systems, your SKU range, your planning cycle

✓

We identify where the forecast bias is introduced - and how to remove it

✓

You leave with a clear Phase 1 outline, whether you engage us or not

✓

No sales script. No deck. A direct commercial data conversation.