# Forecasting *** ### How Moselle builds your forecast Moselle uses machine learning to generate SKU-level demand projections across a 12-month horizon. Here's what's happening under the surface. **It starts with your sales history**\ Mo analyzes your historical order and sales data to identify baseline demand patterns — how much of each SKU sells in a typical week, how that shifts across seasons, and how different channels perform relative to each other. **It accounts for seasonality and trends**\ Mo separates recurring seasonal patterns (a product that always spikes in November) from actual trend changes (a product that's genuinely growing or declining). These are different signals, and treating them the same is one of the most common forecasting mistakes. **It flags anomalies**\ Unusual demand events — a viral moment, a stockout that suppressed real demand, a one-time bulk order — can distort a forecast if they're treated as normal. Mo detects these and adjusts so they don't inflate or deflate your baseline. **It updates continuously**\ Your forecast isn't static. As new sales data comes in, Mo recalculates. If demand shifts, your projections shift with it — without you having to manually revise anything. *** **Two ways to look at your forecast** Moselle supports both **bottom-up** and **top-down** planning views: * **Bottom-up** builds from the SKU level upward — useful for operational decisions like replenishment and production scheduling, where SKU-level accuracy matters. * **Top-down** starts from an overall revenue or category target and distributes it down — useful for financial planning and budgeting. You can use both views and compare them. Where they diverge is usually worth a closer look. *** **A note on forecast accuracy** Moselle gets you 80% of the way there on day one — a working forecast grounded in your actual data, without any manual setup or formula-building. That 80% is enough to make better decisions than most brands are making today. The remaining 20% is where your judgment comes in. Mo's forecast is built from signals it can see — historical sales, seasonality, channel trends. It doesn't know about the promotion you're planning next month, the supplier delay you just heard about, or the new product launch that's going to shift demand in a category. That context is yours to add. You can see how accurate Mo's projections have been for any SKU or category. If a forecast consistently misses in one direction, that's a signal worth investigating — either there's context Mo doesn't have, or there's a pattern in your demand that hasn't been captured yet. Override when you have information Mo doesn't. Add context, adjust the projection, and Mo will recalibrate. Over time, as Mo learns your business, the gap between the forecast and reality narrows. *** ### Why Forecasting Matters **Generate Revenue** — Inventory availability when customers want to buy. Stockouts mean lost sales to competitors. **Reduce Costs** — Avoid overstocking that ties up cash and increases carrying costs. **Catch Mistakes Early** — Surface anomalies before they become costly problems. *** ### Demand Planning vs. Forecasting **Forecasting** is the analytical process of predicting future demand — the numbers. **Demand Planning** is the broader operational discipline that uses forecasts to make inventory decisions, coordinate with suppliers, and align cross-functional teams. Forecasting is an input to demand planning. Moselle handles both: generating forecasts automatically and translating them into actionable replenishment and production plans. *** ### Measuring Accuracy with MAPE **MAPE (Mean Absolute Percentage Error)** measures the average percentage difference between forecasted and actual values. | MAPE | Interpretation | | ------ | ----------------- | | < 10% | Highly accurate | | 10-20% | Good | | 20-30% | Reasonable | | > 30% | Needs improvement | Moselle tracks MAPE automatically by SKU, channel, and time period. *** ### Top-Down vs. Bottom-Up Forecasting #### Top-Down Starts with revenue goals and works backward to unit requirements. * Set revenue target → calculate each SKU's contribution → determine units needed * Best for: new brands, stable demand patterns, financial alignment #### Bottom-Up Analyzes each SKU individually and aggregates upward. * Analyze historical sales per SKU → project demand → aggregate totals * Best for: rich historical data, seasonal products, complex catalogs #### Recommended Approach Use both. Build a top-down forecast from revenue goals, generate a bottom-up forecast from SKU data, compare, and reconcile. Moselle automates this process. --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://learn.moselle.io/planning-and-execution/forecasting.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.