Demand Planning &
Inventory Forecasting FAQ

Creating a 12 to 18 month rolling sales forecast manually requires several steps. First, gather your historical sales data—ideally 2-3 years of weekly or monthly data broken down by SKU and channel. Export this from your WMS, e-commerce platform, or ERP into a spreadsheet.

Next, calculate baseline demand by averaging historical sales while accounting for any anomalies like one-time promotions or stockout periods that would skew your numbers. Apply a trend analysis to identify whether sales are growing, declining, or flat. Most companies use simple linear regression or moving averages for this.

For the rolling aspect, you'll need to update your forecast weekly or monthly by adding new actual sales data and extending your forecast window forward. This means maintaining complex spreadsheet formulas that recalculate automatically, or manually updating ranges each period.

The challenge is that spreadsheet-based rolling forecasts break easily—formulas get overwritten, version control becomes a nightmare with multiple team members, and the process typically takes 4-8 hours per update cycle.

Planster automates this entire process. Once you connect your sales data sources, Planster continuously generates and updates rolling forecasts using statistical models calibrated to your specific demand patterns. What takes hours in spreadsheets happens automatically in Planster, and your forecast is always current without manual intervention.

A predictive purchase ordering plan tells you what to order, how much to order, and when to place each order based on forecasted demand rather than reactive reordering. Building one manually involves several components.

Start by calculating your demand forecast for each SKU over your planning horizon. Then factor in your supplier lead times—how long from PO submission to goods arriving in your warehouse. Add your safety stock buffer to account for demand variability and supply uncertainty.

The formula is: Reorder Point = (Average Daily Demand × Lead Time) + Safety Stock. When your inventory position (on-hand plus on-order minus committed) drops below this point, trigger a purchase order.

For order quantities, you'll want to balance ordering costs against carrying costs. The Economic Order Quantity (EOQ) formula helps: EOQ = √(2 × Annual Demand × Order Cost / Holding Cost per Unit).

The manual challenge is maintaining this across hundreds or thousands of SKUs, each with different lead times, demand patterns, and supplier constraints. Most teams end up with massive spreadsheets that require constant maintenance and still result in stockouts or overstock situations.

Planster calculates reorder points and optimal order quantities automatically for every SKU. It monitors your inventory positions in real-time against your forecast and surfaces exactly which purchase orders need to go out this week, prioritized by urgency. You can place orders directly from Planster and track them through delivery.

The ideal inventory level balances two risks: holding too much inventory (which ties up cash and incurs carrying costs) versus holding too little (which causes stockouts and lost sales). The optimal amount depends on several factors specific to each SKU.

To calculate ideal inventory manually, you need: your average daily sales rate, supplier lead time, demand variability (standard deviation of sales), and your desired service level (what percentage of demand you want to fulfill from stock).

The formula combines cycle stock (average inventory needed to meet demand between orders) plus safety stock (buffer for variability): Ideal Inventory = (Daily Demand × Lead Time) + (Z-score × Standard Deviation × √Lead Time).

The Z-score corresponds to your service level—1.65 for 95% service level, 2.33 for 99%. Higher service levels require more safety stock.

You'll also need to account for: minimum order quantities from suppliers, quantity discounts that might justify holding more, storage constraints, seasonality that affects demand patterns, and product shelf life for perishable goods.

Doing this calculation manually for every SKU and keeping it updated as demand patterns change is extremely time-consuming. Most companies either over-simplify (holding flat weeks of supply) or ignore the calculation entirely and order reactively.

Planster calculates optimal inventory levels for each SKU based on your actual demand variability and lead times. It updates these recommendations continuously as your demand patterns evolve, so you're always holding the right amount—not too much, not too little.

Planning for seasonality requires identifying recurring demand patterns in your historical data and building those patterns into your forecast. Here's how to approach it manually.

First, gather at least 2-3 years of historical sales data to identify patterns. Calculate seasonality indices by comparing each period's sales to the overall average. For example, if December typically sees 150% of average monthly sales, your December seasonality index is 1.5.

The calculation: Seasonality Index = (Period Sales / Overall Average Sales). Do this for each month or week across your historical data, then average the indices for each period across years.

To apply seasonality to your forecast: Forecast = Base Demand × Seasonality Index. If your base monthly demand is 1,000 units and December's index is 1.5, forecast 1,500 units for December.

For CPG brands, you'll likely see patterns around holidays, back-to-school, summer, and gift-giving seasons. But different SKUs often have different seasonal patterns—sunscreen peaks in summer while cold remedies peak in winter.

The manual challenge is maintaining seasonality indices for potentially hundreds of SKUs, updating them as patterns shift, and correctly timing inventory builds ahead of peak seasons (accounting for lead times).

Planster automatically detects seasonality patterns in your data at the SKU level. It identifies when each product's demand typically rises and falls, builds those patterns into your forecast, and alerts you when it's time to place larger orders ahead of seasonal peaks—accounting for your supplier lead times.

Tracking forecast accuracy at the channel level—separating your D2C, Amazon, wholesale, and retail performance—helps you understand where your forecasting works well and where it needs improvement. Here's the manual approach.

First, structure your data so forecasts and actuals are recorded by channel. You'll need columns for: SKU, channel, forecast period, forecasted quantity, and actual quantity sold.

The most common accuracy metrics are:

Mean Absolute Percentage Error (MAPE): Average of |Actual - Forecast| / Actual × 100%. Lower is better. MAPE of 20% means your forecasts are off by 20% on average.

Weighted MAPE: Same calculation but weighted by volume, so high-volume SKUs matter more than slow movers.

Bias: (Sum of Actuals - Sum of Forecasts) / Sum of Actuals. Shows if you're systematically over or under-forecasting. Positive bias means you're under-forecasting.

To track by channel, calculate these metrics separately for each channel. Create a dashboard that shows accuracy trending over time so you can spot when a channel's forecast is degrading.

The manual challenge is maintaining this reporting infrastructure in spreadsheets—pulling actual sales data each period, matching it to historical forecasts, calculating metrics, and visualizing trends. Most teams either skip this entirely or do it quarterly at best.

Planster tracks forecast accuracy automatically at whatever granularity you need—by channel, by SKU, by product category. You can see accuracy metrics in real-time, identify which areas of your business need forecast tuning, and track whether your accuracy is improving over time.

A forecast accuracy report compares what you predicted would sell against what actually sold, helping you measure and improve your forecasting process. Here's how to build one manually.

Start by organizing your data with these columns: SKU, Period, Channel, Forecasted Units, Actual Units Sold. Ensure you're comparing forecasts made at a consistent lead time—comparing a forecast made 8 weeks out to one made 2 weeks out isn't apples-to-apples.

Calculate error metrics for each line: Absolute Error = |Actual - Forecast|, Percentage Error = Absolute Error / Actual × 100%, and Error Direction (over-forecast vs under-forecast).

Aggregate these into summary metrics: MAPE (Mean Absolute Percentage Error) shows average accuracy. Bias shows systematic over or under-forecasting. Root Mean Square Error (RMSE) penalizes large errors more heavily.

Build your report with sections showing: overall accuracy metrics, accuracy by product category, accuracy by channel, accuracy by forecast horizon (1 week out vs 8 weeks out), and trend over time (is accuracy improving or degrading?).

Include a list of top offenders—SKUs with the worst accuracy that deserve investigation. Often you'll find forecast errors cluster around new products, promotional items, or specific channels.

The manual challenge is that this report requires significant data manipulation each period, matching forecast snapshots to eventual actuals and maintaining historical accuracy trends. Most teams find it's 2-4 hours of work per reporting period.

Planster generates forecast accuracy reports automatically. You can view accuracy at any level of detail, see how it's trending over time, and quickly identify which SKUs or channels need attention—all without manual data wrangling.

The best statistical model for your sales data depends on your demand patterns, data history, and business characteristics. Here's a framework for choosing.

For stable demand with no strong trend or seasonality, simple models work well: Moving Averages (average of last N periods) or Exponential Smoothing (weighted average giving more weight to recent data). These are easy to implement in spreadsheets.

For data with clear trends (consistent growth or decline), use Holt's Linear Exponential Smoothing or Linear Regression. These models capture the underlying trend and project it forward.

For seasonal data, Holt-Winters (Triple Exponential Smoothing) handles both trend and seasonality. SARIMA (Seasonal ARIMA) is more sophisticated and can capture complex seasonal patterns.

For highly variable or lumpy demand (common with wholesale or B2B), Croston's Method is designed specifically for intermittent demand patterns where you have many zero-sales periods.

Model selection process: Plot your data visually first—is there obvious seasonality? A clear trend? Then test multiple models against historical data using a holdout period and compare accuracy metrics. The model with the lowest MAPE on your holdout data is typically your best choice.

The manual challenge is that implementing these models in spreadsheets ranges from tedious (moving averages) to nearly impossible (SARIMA). Different SKUs often need different models, and you'd need to re-evaluate as patterns change.

Planster automatically tests multiple statistical models against each SKU's historical data and selects the best-fit model. As your demand patterns evolve, Planster re-evaluates and switches models if needed—no statistical expertise required on your end.

Tracking retail demand separately from D2C and other channels is essential because retail has unique characteristics: longer lead times, promotional calendars, retailer-specific requirements, and often lumpier order patterns. Here's how to approach it manually.

First, isolate your retail sales data by retailer and SKU. Retail data often comes from different sources than D2C—EDI feeds, retailer portals, or distributor reports—so you may need to consolidate from multiple systems.

Build a separate forecasting model for retail that accounts for: retailer order patterns (weekly, bi-weekly, monthly), promotional events and features, retailer-specific seasonality (which may differ from your D2C patterns), new store distributions and lost distributions, and retailer inventory positions if available.

For retail-specific metrics, track: fill rate (what percentage of retailer orders you shipped complete), on-shelf availability if you have access to POS data, weeks of supply at retail, and promotional lift factors for each retailer.

Create a retail demand dashboard that shows: forecasted orders by retailer, upcoming promotional events and their expected lift, inventory positions relative to retail demand, and alerts for potential retail stockouts.

The manual challenge is that retail adds significant complexity—each retailer may require different data formats, have different promotional calendars, and order on different schedules. Managing this in spreadsheets across multiple retailers quickly becomes unwieldy.

Planster lets you segment your demand by channel, including separate retail tracking. You can view retail forecasts, monitor fill rates, and plan inventory specifically for retail demand alongside your other channels—all in one system.

Getting from raw data to automated purchase order recommendations requires connecting multiple data sources and building logic on top. Here's the manual approach.

You need three types of data flowing together: sales/demand data (historical and forecasted), inventory positions (what you have on hand and on order), and supplier information (lead times, minimum order quantities, costs).

For manual integration, most teams export data from each system into spreadsheets: sales reports from Shopify/Amazon/your WMS, inventory snapshots from your WMS or 3PL, and supplier details maintained manually or exported from your ERP.

Build your recommendation logic: calculate reorder points (when to order), economic order quantities (how much to order), and factor in constraints like MOQs, container quantities, and cash flow limits. Cross-reference current inventory positions against reorder points to identify what needs ordering now.

The output should be a prioritized list showing: SKU, supplier, current inventory, forecasted demand, reorder point, recommended order quantity, and order urgency.

The manual challenges are significant: data exports go stale quickly, formula errors cause wrong recommendations, multiple team members working in the same spreadsheet causes version conflicts, and the time to update everything means you're always working with slightly outdated information.

Planster integrates directly with 100+ WMS platforms, 3PLs, and e-commerce tools. Your sales and inventory data flows in automatically—no manual exports or uploads. Planster applies your supplier lead times and constraints to generate purchase order recommendations that are always based on current data. Many teams go from hours of weekly spreadsheet work to simply reviewing and approving the orders Planster surfaces.

Lowering carrying costs while maintaining service levels requires optimizing your inventory—holding the right amount of the right products at the right time. Here's a systematic manual approach.

First, calculate your true carrying costs. Most companies underestimate these. Include: cost of capital (what you'd earn investing that cash elsewhere, typically 8-15%), warehouse storage costs (rent, utilities, labor), insurance, shrinkage and obsolescence, and handling costs. Total carrying cost is often 20-35% of inventory value annually.

Segment your inventory using ABC analysis: A items (top 20% of SKUs driving 80% of revenue) deserve tight management and higher service levels. C items (bottom 50% of SKUs driving 5% of revenue) can tolerate lower service levels and less safety stock.

Reduce safety stock intelligently: Calculate actual demand variability for each SKU rather than using blanket "weeks of supply" rules. Many SKUs have stable demand and need minimal safety stock. Apply higher safety stock only to A items and highly variable SKUs.

Improve supplier relationships: Shorter lead times mean less pipeline inventory. More frequent deliveries mean lower cycle stock. Better supplier reliability means less safety stock needed.

Clean out slow-moving inventory: Identify SKUs with more than 6 months of supply on hand. Discount, bundle, or liquidate to free up cash. Stop reordering items that aren't selling.

The manual challenge is doing this analysis across your entire catalog and keeping it current. Inventory optimization is a continuous process, not a one-time exercise, and most teams lack the bandwidth for ongoing maintenance.

Planster calculates optimal inventory levels for each SKU based on actual demand patterns, lead times, and your service level targets. It identifies overstock situations before they become problems and helps you right-size safety stock—freeing up cash while maintaining the service levels that matter for your A items.

Spreadsheets work early on, but as SKUs, channels, and warehouses increase, they become fragile and time-consuming to maintain. Formulas break, version control becomes chaotic, and forecasting accuracy suffers. As complexity grows, automated forecasting systems like Planster provide more reliable, scalable demand planning without manual upkeep.

Brands typically outgrow spreadsheets when forecasting updates take several hours per cycle, stockouts become frequent, or inventory levels feel reactive instead of planned. If you're managing multi-channel sales or hundreds of SKUs, forecasting software like Planster helps automate rolling forecasts and reorder recommendations so planning stays proactive.

Avoiding stockouts while protecting cash flow requires accurate demand forecasts, correct safety stock levels, and proper reorder timing. Spreadsheet methods often rely on static weeks-of-supply rules that don't adapt to variability. Planster calculates optimal reorder points and safety stock dynamically, helping you hold the right amount at the right time.

The best demand forecasting software supports D2C, Amazon, wholesale, and retail channels in one system. It should account for seasonality, lead times, and channel-specific variability while providing automated purchase order recommendations. Planster is designed specifically for growing e-commerce brands that need multi-channel forecasting without enterprise complexity.

Excess inventory ties up capital that could be reinvested in growth. Accurate forecasting reduces overstock while maintaining service levels, freeing up working capital and improving financial flexibility. By aligning reorder timing and quantities to real demand patterns, Planster helps brands improve cash flow while reducing inventory risk.