share
Forecasting is the heart of supply chain operations — particularly in FMCG, where demand variability, consumer behaviour, and promotional cycles are dynamic and intense. While traditional forecasting models rely on historical sales data, that alone often fails to capture the real-world context behind shifts in demand.
This is where multi-variate forecasting becomes essential — an approach that integrates additional internal or external drivers to improve forecast accuracy and make planning more responsive to reality.
While there are many variables that can influence demand, let’s look at some of the most seen demand drivers across FMCG companies. These drivers, when modelled effectively, can significantly enhance forecast precision and provide better control over planning outcomes.
1. Promotions and Schemes: Internal Yet Unstable Among the most influential internal levers in FMCG is promotional activity. Trade promotions, retail schemes, and price reductions directly drive demand, across channels especially in general trade (GT), modern trade (MT) and E-commerce.
Best Practices:
- Work with finance/marketing teams to create a frozen promotion window — e.g., only include confirmed promotions 2–3 months ahead.
- Design hierarchical fallback logic if SKU-level promos are missing — aggregate to brand/category and apply it proportionally to individual items.
- Some forecasting systems offer separate "event modelling" modules to handle promo data as categorical or numeric inputs, helping planners isolate uplift accurately. This is a better way of working rather than crowding the core statistical engine with lot of layers.
2. Weather: An External Driver With Geographical Challenges: Weather has a well-documented impact on product sales — especially in categories like skincare, beverages, and personal care. Rainfall, humidity, and temperature changes can significantly influence short-term demand.
Best Practices:
Since weather is not a fixed input but a forecast in itself, it's important to test its impact by comparing statistical models both with and without weather variables. This helps confirm if its inclusion genuinely improves forecast accuracy.
3. MRP as a Demand Driver in Price-Sensitive Markets: While MRP (Maximum Retail Price) is often viewed as a static attribute, it can act as a powerful demand change signal, especially in price-sensitive categories like soaps, creams, and oral care. Changes in MRP can lead to abrupt demand shifts—either upward or downward—based on perceived value and consumer affordability. These shifts, known as demand step changes, often occur at specific price thresholds where consumer psychology is sensitive. Recognizing these price-elastic points is critical for forecasting accuracy.
Best Practices:
- Combine MRP shifts with promotion activity to isolate true price elasticity.
- Track price changes at SKU-channel-level wherever possible
4. Distribution Drives: Reach Fuels Demand: Distribution drives refer to focused initiatives aimed at expanding the physical or digital availability of products in the market. These initiatives can target increasing the reach of a brand by onboarding new distributors, penetrating untapped outlets, or enhancing product visibility at the point of sale. In FMCG, where availability is directly tied to sales performance, a well-executed distribution drive can act as a powerful demand enabler.
Best Practices:
- Track drives as event flags in a structured event calendar. This will help planners in making fine tunings to forecast as the effects may or may not have the same effect in the future.
- Understand historical uplift impact at different regions and estimate future impact.
5. Media Spend (TV, Digital, Social): Media exposure directly impacts brand recall and short-term sales, especially for personal care and impulse categories. TV, OTT, and influencer campaigns can show spikes in demand during or right after campaign windows.
Best Practices:
- Proper data collection segregated by category- TV, Digital, Print, Outdoor,etc.
- Geographical Tagging – Strategize on how to map it to regions thereby aligning with our forecasting levels.
6. Exception: Fill Rate Loss as a Sales Correction Tool:
Fill rate loss is often misclassified as a forecast input — but it's not a forward-looking driver. Instead, it’s a lag indicator of supply or planning failure, and its only practical use is in correcting historical demand.
If a customer ordered 1000 units but only 600 were fulfilled due to stock-outs, the system would record only 600 — underreporting actual demand by 40%.
Best Practices:
Use fill rate-based sales corrections selectively and only when the impact on forecast accuracy is significant. If stockouts are sporadic or short-lived, their influence on statistical forecasts is often negligible. However, for products with prolonged stockouts, failing to account for lost sales can distort demand signals. In such cases, correcting sales history may be justified — but only if the value outweighs the effort.
Forecasting in FMCG is no longer just about projecting numbers — it’s about understanding the context behind the numbers. Multi-variate forecasting adds this context, but it comes at a cost — of data, process, and cross-functional alignment. Not every data point is worth the effort. The key is to build a prioritization model that blends business intuition with data readiness.
Start small. Test the uplift. Automate where possible. And most importantly, collaborate — because forecasting is not a function, it’s an orchestration of data, people, and strategy.