Forecast Fragmentation in FMCG: Why Demand Planning is Breaking Down—And How to Fix It

The Growing Chaos Behind the Shelf

In today’s fast-moving consumer goods (FMCG) industry, demand planning is no longer just a forecast—it’s a battleground.

As global supply chains stretch thinner, product lifecycles shrink, and consumer expectations rise, many FMCG firms are grappling with a frustrating paradox: more data, yet less clarity resulting in Forecast fragmentation—a state where multiple versions of “truth” emerge across commercial, marketing, and supply chain teams, each based on conflicting assumptions, data sets, or objectives.

For planners, this isn’t just noise. It’s a supply chain liability.

Fragmentation Across Markets: One Problem, Many Faces

Forecast fragmentation wears different faces around the globe:

• LATAM: Promotions dominate demand. A 2-for-1 discount in Brazil can triple weekly volumes—yet models often can’t detect or adjust for these spikes in time.
• Southeast Asia: Channel volatility rules. An unplanned TikTok flash sale can spike demand in Indonesia, while traditional outlets see flatlining volumes.
• Europe: Retailer power is king. Hardline order caps from discounters make it challenging to forecast replenishment demand with confidence
• North America: SKU proliferation continues. From oat milk variants to organic pet food, planners face an avalanche of slow-moving, erratic SKUs.

In most cases, the same planning approach doesn’t work, resulting in different forecast variations that no one trusts.

The Data Overload Dilemma

FMCG companies have no shortage of data—POS, syndicated retail, digital shelf, weather, social trends, even competitor pricing.

However, most planning models rely heavily on internal sales history, with little weight given to these external variables. Even when advanced ML models are introduced, they often falter due to:

• Poor SKU segmentation (treating all products the same),
• Lack of real-time signal integration (external data is stale by the time it’s used), and
• Opaque models that planners don’t trust or understand.

We need to stop treating all SKUs as the same. Forecasting is not rocket science. Segmenting and personalising your models will generate a better result.

From Accuracy to Actionability

For years, the demand planning conversation has centered on accuracy—MAPE, bias, CoV. In the real world of FMCG execution, that mindset is breaking down.

Why? Because:

• Accuracy ≠ usability. A model might be 92% accurate in aggregate but wildly wrong at the SKU or regional level.
• Precision is useless if the supply can’t follow. In highly constrained supply environments, responsiveness matters more than precision.
• Time-to-action is the new KPI. If a forecast takes two weeks to approve, you’re already too late.

We need to evolve from forecast accuracy obsession to forecast usability- the goal is aligned, actionable plans, not perfect numbers.

Bridging the Gap with Supply Planning

Even the best forecasts break if the supply network can’t flex. The critical bottlenecks today aren’t just in accuracy—they’re in translation:

• Supply lead times are longer than planning cycles. This means buffers must be proactive, not reactive.
• Over-precision in demand plans leads to rigidity in supply plans, increasing bullwhip effects.
• Shared capacity across co-packers, CMOs, and regional plants makes it harder to lock firm builds without flexibility.

Plan for error, include buffers, cross-train suppliers and enable agile replanning for better forecasting

Strategies to Un-fragment Demand Planning

Here’s how leading FMCG firms are breaking the fragmentation trap:

1. Forecast by Behavior, Not Just History

Segment SKUs into behavioral clusters:

• Smooth: Base forecast with seasonality
• Lumpy: Use Poisson models + min/max triggers
• Erratic: Rely on qualitative overrides or supply buffering

Why it matters: Tailored models avoid overfitting and reduce planner workload on the long tail.

2. Integrate External Signals at the Right Level

Bring in weather, macro trends, search volumes—but don’t overload SKU-level models. Use them for:

• Baseline shaping: e.g., rainfall driving soup demand
• Promo modeling: e.g., competitor price drops
• Scenario planning: e.g., recession-adjusted volumes

Why it matters: Planners don’t need a perfect answer—just better early signals.

3. Enable Cross-Functional Consensus Forecasting

Move from siloed forecasts to true S&OE (Sales & Operations Execution):

• Weekly consensus touchpoints across sales, marketing, and supply
• Real-time adjustment thresholds (e.g., “auto-approve up to +10% swing”)
• Root cause review of past forecast gaps

Why it matters: A slightly less accurate shared forecast is better than a hyper-accurate disputed one.

4. Invest in Demand-Supply Synchronization, Not Just Planning Tech

Instead of just adding another AI layer, focus on:

• Responsive supplier relationships
• Agile production plans
• Lead time buffers at regional warehouses

Why it matters: Flexibility trumps precision. The best forecasts fail gracefully.

The Real KPI: Confidence in the Forecast

When planners, marketers, and supply managers don’t trust the demand plan, everything breaks down:

• Safety stocks go up (cost),
• Expedites increase (chaos),
• Service drops (customer impact),
• And planners burn out (turnover).

The most valuable output of a demand planning system today is not a number—it’s confidence.

Closing Thoughts

Forecast fragmentation is not a technology problem—it’s a collaboration and adaptability challenge.

As demand becomes more volatile and global supply chains remain strained, the goal is no longer perfection. It's resilience.

FMCG companies that succeed in the next decade will be those that:

• Treat demand planning as a team sport, not a back-office function,
• Segment their strategies by SKU behavior and market volatility,
• And build systems and processes that flex with the consumer, not just model them.