Many brands approach sales forecasting by first reviewing last year's sales and then adding this year's growth targets: the marketing department estimates incremental sales from campaigns, the sales department compiles customer orders, the channel department relays distributors' inventory needs, and the supply chain arranges production accordingly. All departments have their rationale, but the final number often does not reflect true market demand. This is because sales targets, distributor orders, brand shipments, and sell-through are essentially four different sets of numbers. Sales targets answer what the company wants to achieve; distributor orders reflect how much inventory customers plan to buy; brand shipments record what the company actually delivers; and sell-through is closer to what consumers actually purchase. Orders may include channel stuffing, stockpiling, and promotion pull-forward, and shipment growth may simply be inventory moving to the channel. Mixing these numbers together distorts the forecast from the start. To do sales forecasting properly, brands should first establish an interpretable baseline model: Forecast sales = base sales + growth increment - risk deduction. Base sales are derived from historical sell-through, seasonal patterns, regional and channel differences, repeat purchase performance, and product life cycle. Growth increments come from new outlets, new product launches, festive peak seasons, marketing investment, promotions, and price adjustments. Risk deductions need to account for stockouts, channel inventory, price fluctuations, competitor actions, and terminal execution. Meanwhile, mature products and new products cannot use the same approach. Mature products have historical patterns to follow; the key is cleaning abnormal data and identifying a stable baseline. New products lack historical references, so they require combining similar products, test launch feedback, terminal coverage, early sell-through, and consumer feedback to continuously adjust the sales range. A truly effective forecast should not be a single number; it should simultaneously form three scenarios: baseline, optimistic, and cautious. Each outcome corresponds to clear business assumptions. When conditions change, the forecast should be adjusted accordingly. In reality, however, sales, order, inventory, and channel data are scattered across different departments. Manual consolidation is time-consuming, and forecasts often become outdated as soon as they are completed. AI can help brands quickly consolidate data, clean anomalies, establish sales baselines, identify regional, product, and channel differences, summarize growth drivers and risk variables, generate multi-scenario forecast drafts, and continuously compare actual results against forecast deviations. The brand team remains responsible for judging whether orders are genuine, whether growth can be realized, whether inventory is healthy, and ultimately how to allocate production and channel resources. On September 17-18, 2026, in Zhengzhou, China, the "FMCG Growth AI Bootcamp" will focus on real brand operating scenarios, helping teams apply AI to data consolidation, sales forecasting, inventory judgment, and business review. The goal is to make sales forecasting no longer a one-time form-filling exercise, but a business mechanism that is explainable, collaborative, calibratable, and continuously updated.
Many brands approach sales forecasting by first reviewing last year's sales and then adding this year's growth targets: the marketing department estimates incremental sales from campaigns, the sales department compiles customer orders, the channel department relays distributors' inventory needs, and the supply chain arranges production accordingly.
All departments have their rationale, but the final number often does not reflect true market demand.
This is because sales targets, distributor orders, brand shipments, and sell-through are essentially four different sets of numbers.
Sales targets answer what the company wants to achieve; distributor orders reflect how much inventory customers plan to buy; brand shipments record what the company actually delivers; and sell-through is closer to what consumers actually purchase.
Orders may include channel stuffing, stockpiling, and promotion pull-forward, and shipment growth may simply be inventory moving to the channel. Mixing these numbers together distorts the forecast from the start.
To do sales forecasting properly, brands should first establish an interpretable baseline model: Forecast sales = base sales + growth increment - risk deduction.
Base sales are derived from historical sell-through, seasonal patterns, regional and channel differences, repeat purchase performance, and product life cycle. Growth increments come from new outlets, new product launches, festive peak seasons, marketing investment, promotions, and price adjustments. Risk deductions need to account for stockouts, channel inventory, price fluctuations, competitor actions, and terminal execution.
Meanwhile, mature products and new products cannot use the same approach. Mature products have historical patterns to follow; the key is cleaning abnormal data and identifying a stable baseline. New products lack historical references, so they require combining similar products, test launch feedback, terminal coverage, early sell-through, and consumer feedback to continuously adjust the sales range.
A truly effective forecast should not be a single number; it should simultaneously form three scenarios: baseline, optimistic, and cautious. Each outcome corresponds to clear business assumptions. When conditions change, the forecast should be adjusted accordingly.
In reality, however, sales, order, inventory, and channel data are scattered across different departments. Manual consolidation is time-consuming, and forecasts often become outdated as soon as they are completed.
AI can help brands quickly consolidate data, clean anomalies, establish sales baselines, identify regional, product, and channel differences, summarize growth drivers and risk variables, generate multi-scenario forecast drafts, and continuously compare actual results against forecast deviations. The brand team remains responsible for judging whether orders are genuine, whether growth can be realized, whether inventory is healthy, and ultimately how to allocate production and channel resources.
On September 17-18, 2026, in Zhengzhou, China, the "FMCG Growth AI Bootcamp" will focus on real brand operating scenarios, helping teams apply AI to data consolidation, sales forecasting, inventory judgment, and business review.
The goal is to make sales forecasting no longer a one-time form-filling exercise, but a business mechanism that is explainable, collaborative, calibratable, and continuously updated.
