Spark Inventory Blog

A holiday peak can look obvious in last year's sales chart and still produce the wrong purchase order. The problem is rarely recognizing that demand is seasonal. The harder job is turning that pattern into a per-SKU forecast that accounts for current stock, incoming inventory, supplier lead time, channel mix, and the cash you can commit. That is what seasonal demand forecasting should do: support a buying decision, not just describe a past pattern.

For a growing product business, the consequence of getting it wrong is familiar. Buy too late and a fast seller goes out of stock while demand is strongest. Buy too early or too deeply and inventory occupies warehouse space and working capital long after the peak has passed. A useful forecast gives the operator a clear view of both risks before the reorder is due.

What seasonal demand forecasting needs to answer

Seasonality is a repeatable pattern in demand tied to a period: a holiday, weather shift, retail calendar, back-to-school window, or annual replenishment cycle. It is not every temporary sales spike. A one-time influencer mention, a clearance event, or a stockout recovery can distort the pattern if it is treated as normal seasonal demand.

The operational questions are more specific than, "Will sales rise in November?" A planner needs to know which SKUs will rise, in which warehouse or channel, when the lift begins, how long it lasts, and whether inventory must be ordered before the demand arrives. That requires connecting demand history to supply timing.

For each SKU, a seasonal forecast should inform four decisions: expected demand during the replenishment window, days of supply at the current run rate, stock risk before the next receipt lands, and the reorder recommendation needed to cover the period. The output is a buying plan, not a chart left for someone else to interpret.

Start with a clean demand signal

A forecast is only as useful as the history behind it. Sales data needs context before it becomes a planning input. If an item was out of stock for two weeks last December, recorded sales for those weeks do not represent actual demand. Likewise, a temporary promotion may be a valid planning signal if it will recur, but noise if it will not.

Review historical demand at the level where you purchase and hold inventory. In many businesses that means the SKU and warehouse, with channel detail available for investigation. An aggregate product-family forecast can hide a stockout risk in one color or size, while a channel-only view can miss inventory shared across channels.

The practical checks are straightforward:

• Flag stockout periods, discontinued items, and incomplete sales history so they do not quietly pull the forecast down.
• Separate planned promotions, bundles, and price changes from baseline demand where their future treatment differs.
• Confirm product substitutions and new variants, especially when a predecessor SKU carries most of the usable history.
• Reconcile sales, on-hand, committed, and incoming inventory so the buying plan begins from the same operating reality as the warehouse and purchasing team.

This is also where spreadsheets become fragile. They can calculate a seasonal index, but they often depend on someone remembering which cells were adjusted for a stockout or late receipt. A planning system should retain the assumptions and show why a forecast changed.

Build the forecast around the replenishment window

A seasonal forecast without supplier lead time is incomplete. If a supplier needs 60 days to produce and deliver an item, a peak that starts in late November must affect the buying plan in September, not when sales begin accelerating.

The relevant planning window is usually supplier lead time plus a receiving and review buffer. If you order every 30 days, include the time until the next ordering opportunity as well. Demand during that window is the demand your inventory position must cover.

A simple planning relationship is:

Required inventory position = forecast demand during the protection period + safety stock

Suggested reorder quantity = required inventory position - available inventory position

Available inventory position generally includes on-hand inventory and reliable incoming inventory, then subtracts allocations or committed stock where applicable. The word reliable matters. An incoming purchase order that is past its expected arrival date should not be treated the same as a confirmed shipment due next week.

Safety stock is not a fixed percentage for every SKU. It depends on demand variability, supplier reliability, replenishment flexibility, and the cost of a stockout. A core item with an inconsistent supplier may need more protection than a seasonal accessory that can be replenished quickly. The operator should be able to adjust this policy and see how it changes the recommendation.

Use history carefully when this season is different

Last year is valuable, but it is not a verdict. A business may have added a wholesale account, changed its assortment, expanded to a new warehouse, or shifted promotional timing. Comparing this year's seasonal demand only to last year's total sales can lead to a confident but misleading buy.

A better approach starts with a baseline and then applies explicit adjustments. For example, a planner may use last year's weekly pattern for a mature SKU, adjust for recent run rate, and separately account for a confirmed retail order or planned promotion. The forecast should distinguish between expected recurring demand and known events rather than burying both in one unexplained number.

New products require a different method. There is no meaningful seasonal history for a SKU introduced three months ago. Use analogous items, product-category behavior, early sell-through, and the seasonality of the item it replaced. Keep the assumption visible, because new-product forecasts deserve more frequent review and generally more cautious purchase quantities.

Channel behavior matters too. A product can peak earlier in wholesale than direct-to-consumer sales because retail buyers place orders ahead of consumer demand. If the same inventory pool supports both, the buying plan needs to reserve inventory for confirmed wholesale commitments while forecasting the consumer-facing peak. Treating every sales channel as one undifferentiated history can hide that timing difference.

Turn the forecast into an operating cadence

Seasonal demand forecasting works best when it is reviewed before the business is under pressure. Monthly planning may be adequate for stable, long-lead-time items in the off-season. As a peak approaches, weekly review is often more appropriate for high-value SKUs, constrained suppliers, and items with rising stock risk.

The review should focus on exceptions, not force the team to inspect every SKU equally. Start with items projected to stock out before the next receipt, items whose incoming inventory has moved, and items with excess days of supply after the season. Then check whether the recommended buy respects supplier minimums, case packs, available cash, and warehouse capacity.

This is where a forecast becomes operationally useful. Spark Inventory can combine sales history with on-hand and incoming inventory, supplier lead times, and inventory policies to produce a per-SKU forecast, identify stock risk, and prepare a buying plan. It can also prepare draft purchase orders from approved recommendations. The operator reviews, adjusts, and approves each draft before anything is sent or committed.

That approval step is not administrative friction. It is where commercial judgment belongs. A buyer may know a supplier is capacity-constrained, a customer order is uncertain, or a product refresh will shorten the selling window. The system should prepare the decision and explain its inputs, while the operator remains accountable for the commitment.

Watch for excess after the peak

Stockout prevention gets most of the attention, but seasonal overbuying often creates the longer problem. A reorder that protects December demand may be unnecessary if it arrives after the selling window. The forecast should show expected demand after the seasonal peak, not just the risk before it.

When reviewing a proposed order, ask two questions: Will this quantity arrive in time to serve the peak? And if demand returns to baseline on schedule, how many days of supply will remain? The answer may justify a smaller order, an earlier order, or no order at all. It depends on margin, storage constraints, supplier minimums, and whether the product is sellable next season.

A seasonal buying plan should therefore include an exit view. For each material purchase, examine the expected inventory position at the end of the season and the working capital still tied up. That makes trade-offs visible before the order is placed, when they can still be managed.

The next seasonal peak is rarely won by finding a more elaborate forecasting formula. It is won by reviewing the right SKUs early enough, connecting demand to supply timing, and making each reorder with a clear view of both availability and cash.