Which Approaches Can Be Used to Forecast Sales in a High-Volatility Demand Environment
1. Why Classic Forecasting Methods Fail in an Unstable Environment
In high-volatility demand environments — such as during martial law, economic crises, price shocks, or rapidly changing consumer behavior — traditional forecasting models based on trends and seasonality often break down. A business predicts growth but ends up with a decline. Why? Because such models rely on past data that is no longer a reliable reference. In such cases, we need flexible, adaptive, and layered approaches.
2. What Does “Demand Volatility” Mean in Practice?
It means unpredictable changes in sales volumes:
yesterday – 300 units, today – 90, tomorrow – 600;
external factors like inflation, news, or weather instantly impact behavior;
customers suddenly shift channels, product formats, or preferences.
In such conditions, every forecasting error leads to financial loss — either inventory sits unsold or demand exceeds supply.
3. Key Forecasting Approaches for High-Uncertainty Conditions
3.1. Short- and Ultra-Short-Term Forecasting
Replace 6–12 month forecasts with models for 1–4 weeks. Update data frequently and rebuild models regularly.
Tools include:
rolling forecasts (moving windows);
automatic weekly model updates;
daily dashboards in BI systems.
3.2. Weak Signal Analysis
Use early signals from search trends, website behavior, product views, customer queries, or support requests to detect upcoming demand shifts.
Example:
An increase in search volume for “contactless delivery” before a rise in related orders.
3.3. Scenario Modeling
Instead of a single forecast, create multiple: optimistic, baseline, and pessimistic. Each scenario includes tailored plans for purchasing, logistics, and marketing.
The benefit:
You can switch strategies quickly when conditions change — no panic, no delays.
3.4. Machine Learning-Based Models
These can factor in hundreds of inputs — not just past sales and seasonality, but also exchange rates, competitor activity, fuel prices, customer sentiment, and more.
Examples:
XGBoost, Prophet (Meta), LSTM for time series forecasting.
3.5. Integration with External Data
Pull in data from Google Trends, weather APIs, market indices, social media analytics.
Advantage:
Allows for a broader view beyond internal statistics.
4. Changing the Planning Culture Within the Company
4.1. Forecast Ranges Instead of “One True Number”
No one can predict exact future orders. But you can provide an interval like “1,800 to 2,500 units” — and prepare for both scenarios.
4.2. Incremental Planning
Avoid building an entire plan in one go. Instead, review and revise forecasts every 1–2 weeks as new data comes in.
4.3. Involving Sales and Marketing in Forecasting
Analysts alone can’t predict accurately without real-world input. Collaborative planning with sales and marketing gives much more grounded, flexible results.
5. Real-World Example
A Ukrainian food distributor in 2023 faced extreme volatility: shifting demand, changing delivery routes, unpredictable store schedules. They replaced long-term plans with 5-day rolling forecasts, using data from Google Trends, weather, and order histories.
Results:
product write-offs fell by 34%,
inventory turnover improved by 21%,
replenishment accuracy increased by 27%.
6. How BAT Can Help
BAT enables you to:
build rolling forecasts for 1–4 weeks with daily updates;
integrate external data (weather, search trends, exchange rates);
generate event-driven scenarios with risk scoring;
detect weak signals from behavioral changes;
sync forecasts with purchasing, inventory, logistics, and marketing workflows.
BAT doesn’t replace forecasting — it makes it agile, responsive, and effective, even under pressure.
Conclusion
In a world where demand changes daily, survival depends not on “guessing right” but on adapting fast. A reliable forecast isn’t one fixed number — it’s a living system that reacts to the market. And the more complex the environment, the more vital it is to have a tool that looks ahead. BAT is exactly that tool — giving businesses the insight to stay one step ahead.