Your Production Demand Forecast Is Wrong Because Your Stockouts Suppressed the Signal

4.2 units per store per week. That is what the POS data says a $25M specialty food brand's flagship hot sauce sells at its largest regional retailer. The co-packer schedules production around that number. The brand orders raw materials around that number. The production demand forecast submitted to the S&OP cycle uses that number.

The number is wrong. A three-week out-of-stock event in February (shelves empty while the co-packer waited for packaging materials) dragged the trailing average down by 16%. True demand, corrected for the stockout period, is 5.0 units per store per week. The brand is systematically under-producing, which causes more stockouts, which further suppresses observed velocity, which further lowers the forecast. It is a doom loop, and it is invisible in any demand plan built from raw POS data.

The Structural Error in the Production Demand Forecast

NielsenIQ estimates that out-of-stock rates in specialty food run 8-12% of store-weeks annually. Each stockout period creates a hole in the velocity data. A forecast model trained on data with holes, and no mechanism to identify or fill them, will understate demand proportionally.

The error compounds in co-packer-dependent supply chains because production decisions have long lead times. A typical co-packer requires 6-8 weeks from PO to finished goods. If the demand forecast says 4.2 and true demand is 5.0, the production order covers only 84% of what the shelf will pull. That shortfall lands 8 weeks from now as another stockout, which suppresses velocity further, which makes the next forecast even lower.

McKinsey's research on supply chain planning finds that demand signal accuracy is the single largest driver of inventory optimization: more than lead time reduction, safety stock tuning, or order frequency changes. For a co-packer-dependent brand, signal accuracy starts with correcting for the stockouts that distort the signal.

Correcting Velocity Before Forecasting Changes the Math

The correction method is straightforward. Identify weeks where on-shelf availability dropped below a threshold (typically using store-level inventory or distributor out-of-stock flags). For those weeks, replace observed velocity with an imputed value derived from a seasonal index built on the product's in-stock periods. The corrected velocity series becomes the input to the forecast model.

Lailara's production demand forecast tool applies this correction using STL (seasonal-trend decomposition using Loess) on the in-stock velocity series, then runs a 12-week rolling forecast by SKU. The output pairs the demand projection with a stockout date and a production decision deadline.

The distinction matters. A demand forecast that says "you need 5,000 cases over the next 12 weeks" is useful. A forecast that says "you will stock out in week 9, and the deadline to place the co-packer PO was week 3" is operational. The S&OP process needs the second format, because the decision it governs is not "how much" but "by when."

Co-Packer Constraints Turn a Forecast Into a Production Plan

A demand number without capacity constraints is a wish. Co-packers have minimum run sizes (often 500-1,000 cases per production line), maximum weekly throughput, and scheduling windows that fill weeks in advance. A forecast that ignores these constraints produces orders that cannot be fulfilled on time.

The production demand forecast overlays co-packer capacity and lead-time constraints on the corrected demand signal. For each SKU, it calculates: current inventory, projected depletion rate (from corrected velocity), the date inventory hits safety stock, and the latest date a production order can be placed given the co-packer's lead time. If that deadline has already passed, the tool flags it red.

This connects directly to the financial decisions documented in the co-packer agreement. Minimum order quantities, production line allocation, and rush order surcharges are contract terms. A brand that consistently places orders inside the lead-time window pays 15-25% premiums for expedited production, not because it wanted to rush, but because its demand signal was too low by the time the forecast was built.

The Doom Loop Has a Dollar Figure

For a $25M brand with 50 SKUs, a systematic 16% understatement of demand on even the top ten SKUs translates to roughly $450K-$700K annually in combined costs: lost revenue from stockouts, expediting premiums from late co-packer orders, carrying costs on safety stock built as a hedge against a forecast nobody trusts, and distributor penalties from shorted orders.

These are not visible as a single line item. They are spread across COGS (expediting), revenue (lost sales), and operational overhead (manual replanning, broker calls, retailer apology meetings). The connection between a February out-of-stock event and a June expediting surcharge is invisible without a tool that traces the signal path from POS to production.

Cinderhaven Provisions (a synthetic $25M specialty food brand built for these demonstrations, not a real company) shows this path explicitly. The Artisan Sauce hero SKU (CHP-AS-001) has an observed velocity of 4.2 units per store per week and a corrected velocity of 5.0: a 16% understatement of true demand. The tool traces that gap through the production schedule to a projected stockout in week 9 and a production decision deadline in week 3. If the deadline passes without a PO, the tool shows the cost: rush production surcharge plus lost revenue during the stockout window.

The Fix Is Correcting the Signal, Not Adding Safety Stock

Most brands respond to forecast inaccuracy by padding safety stock: carrying an extra two to four weeks of inventory as a buffer against the forecast being wrong. This costs money (carrying costs run 20-30% of inventory value annually) and does not fix the underlying problem. The forecast is still wrong. The buffer just delays the consequences.

Correcting the velocity signal before forecasting breaks the doom loop at its source. The forecast improves. Production orders align with actual demand. Co-packer POs go out inside the contracted lead time instead of as rush orders. Stockout frequency drops. The velocity data for the next planning cycle is cleaner because fewer stockout periods distort it. Each cycle reinforces accuracy rather than compounding error.

Get the Forecast That Accounts for Your Stockouts

Lailara's demand forecast and production planning engagement starts with your POS data and co-packer contracts, corrects the velocity signal for out-of-stock suppression, and delivers a rolling 12-week production plan with decision deadlines per SKU. The forecast you are running today has a structural bias. Now you know where it comes from.