How to Plan Promotions Around Fulfillment Capacity

Most promotion plans start in the wrong place.

Merchandising sets the sales target. Marketing builds the campaign. Finance signs off on the margin. Then, somewhere toward the end of the process, someone asks the fulfillment team: “Can we handle this?”

By that point, the answer is expected to be yes.

That sequence creates a predictable problem. The promotion does exactly what it was designed to do and generates demand, but the inventory position or fulfillment network cannot support it. Orders pile up. Core sizes disappear. Split shipments increase. The DC adds overtime. Delivery promises slip. What looked like a strong promotion on the sales report becomes considerably less attractive once the operational cost shows up.

Fulfillment capacity should be a planning constraint from the start, alongside inventory, margin and demand. Not an operations problem to solve after the promotional calendar has been approved.

Start With Promotional Demand, Not the Normal Sales Forecast

A normal demand forecast is not a promotional forecast.

The distinction sounds obvious, but it gets blurred surprisingly often. A planner takes the current forecast, applies an uplift percentage based on last year's event, and calls the job done.

That can be directionally useful. It is not enough to commit inventory and fulfillment capacity.

Promotional demand needs to be thought about in two pieces:

Baseline demand + incremental promotional demand = expected total demand

The baseline matters because those sales would probably have happened anyway. The promotional component is what the event is adding, shifting or accelerating.

Oracle's retail forecasting methodology treats promotions and other events as causal effects on demand rather than simply part of the normal sales pattern. That distinction matters because the same discount can behave very differently under different conditions.

Say a retailer ran 25% off outerwear last October and wants to repeat it this year. Last year's uplift is useful evidence, but I would not copy it straight into this year's plan.

Was the assortment equally mature? Did the campaign get the same marketing exposure? Is traffic running ahead or behind last year? Were the hero SKUs fully available? Has the channel mix shifted toward ecommerce? Is the promotion landing before a weather change instead of after it?

One of the bigger traps is censored demand.

If a hero SKU stocked out halfway through last year's promotion, its sales history tells you how many units you were able to sell. It does not tell you how many customers wanted to buy it. Using those sales as the starting point for the next event can bake the previous stockout into the new forecast. Shopify makes the same point around stockouts obscuring true demand.

This is why the output of promotion planning cannot stop at “we expect $1.5 million in sales.”

The operation cannot pick revenue.

You need expected units and orders by day, SKU, channel and fulfillment location where possible. That's the level where the forecast starts becoming useful for inventory, allocation and capacity decisions.

Translate the Promotion Forecast Into Actual Fulfillment Workload

Merchandising and operations often talk about the same promotion in different units.

The merchant sees units and revenue. The DC sees orders, lines, picks, cartons and shipments.

You need to connect the two:

Units sold → orders → order lines → picks → packs/cartons → shipments

Twenty thousand promotional units tells the fulfillment team very little on its own.

Twenty thousand units spread across 18,000 mostly single-item ecommerce orders is one workload. Twenty thousand units sold through 6,000 larger baskets is another. Same unit demand. Completely different pick-pack profile.

Assortment matters too.

A promotion concentrated around six hero SKUs can generate heavy volume while remaining relatively straightforward to pick. A broad “25% off everything” event might generate the same sales but spread picks across hundreds or thousands of SKUs. That means more travel, more replenishment into pick faces, more opportunities for inventory discrepancies and potentially more split orders.

Identify the real capacity constraint

“Warehouse capacity” is not one number.

DHL's peak-season guidance points to warehousing, transportation, inventory handling and labor as capacity considerations during demand surges. In practice, the constraint could be even more specific.

Maybe there are enough pickers but not enough packing stations. Maybe packing is fine, but replenishment cannot keep fast-moving SKUs in forward pick locations. Maybe the DC can process the volume but the carrier collection cannot take another 4,000 parcels before cutoff.

This is where theoretical capacity gets retailers into trouble.

A DC may technically be capable of processing 25,000 orders per day. But perhaps that number came from a peak shift with full attendance, a favorable SKU mix and additional carrier collections.

If sustainable throughput under the actual promotion plan is closer to 19,000, then 19,000 is the number I would plan against.

Do the comparison step by step. Forecast the workload through picking, replenishment, packing, sortation, carrier handoff and any other meaningful constraint in the network.

The lowest sustainable capacity is the constraint.

Promotional planning has to respect it.

Test Inventory and Allocation at SKU and Location Level

A retailer can have enough inventory for a promotion in total and still have the wrong inventory almost everywhere that matters.

This happens constantly in apparel and footwear.

Suppose a style has 8,000 units available and the promotional forecast calls for 6,000. At style level, the position looks comfortable.

Then you look at the size curve.

Medium and large are already running tight. XS and XXL are carrying excess WOS. The aggregate unit position hides the fact that the commercial sizes are likely to break early in the event.

Customers do not buy “style inventory.” They buy the specific size and color they want.

Once M and L disappear, you have a stockout from the customer's perspective even if the system still shows thousands of units available. The promotion can then leave you with exactly the inventory you did not want: broken size curves and residual units in peripheral sizes that are harder to clear without another markdown.

That is why promotion planning needs to test demand against available-to-promise inventory, inbound receipts, safety stock, replenishment timing and allocation at the level where decisions actually happen.

Oracle explicitly connects forecasting to replenishment, purchasing and allocation. It also accounts for out-of-stocks and promotional effects as part of retail demand forecasting.

Look forward as well.

A promotion that produces excellent sell-through can still leave the business in a bad inventory position. What happens to WOS when the event ends? Are core replenishment styles being pushed below a sensible coverage level? Will the next receipt arrive before you stock out? Are slow colors or sizes still sitting at eight weeks of supply while the best sellers fall below one?

Allocation adds another layer.

You might have enough inventory across the network but not in the nodes expected to fulfill the demand. That can turn what should have been a clean order into a split shipment, transfer or stockout.

Rebalancing inventory before the event is usually cheaper and cleaner than discovering the allocation problem while order volume is already spiking.

This is also where forward-looking inventory monitoring earns its keep. A platform like Flagship can help planners see projected SKU and size-level inventory positions before a promotion lands, rather than finding the problem afterward in a spreadsheet full of yesterday's sales.

Use Capacity Scenarios to Decide Whether the Promotion Needs to Change

I would not approve a significant promotion against one demand number.

Promotional forecasts are uncertain by definition. Marketing response varies. Weather changes. Competitors move. A product suddenly gets traction on social. Even when the underlying forecast is good, there is a range of plausible outcomes.

Build at least three scenarios:

1. Expected demand: the outcome the business is planning around.
2. Upside demand: the promotion materially outperforms expectations.
3. Downside demand: response is weaker than planned.

For each scenario, translate the sales forecast into inventory consumption and fulfillment workload.

The upside case deserves particular attention because retailers have a habit of treating upside as automatically good news.

Commercially, yes. Operationally, maybe.

If the promotion generates 30% more orders than expected and creates a two-day backlog, you now have to ask what that incremental revenue actually cost. Overtime goes up. Carrier options get more expensive. Customer service contacts rise. Orders may be cancelled. Full-price demand after the event may hit depleted inventory.

Define responses before launch instead of improvising them after the dashboard turns red.

The expected scenario might fit within normal shifts plus scheduled overtime. The upside case could trigger another labor shift or carrier collection. Beyond a defined threshold, the business might extend delivery promises or reduce further promotional exposure.

Shape demand instead of endlessly adding capacity

This is the part retailers sometimes resist.

If demand is likely to exceed practical capacity, you do not always need more capacity. Sometimes you need to change the promotion.

DHL specifically recommends timing promotional activity so demand is generated when sufficient delivery capacity is available.

There are plenty of ways to do it without killing the event.

Stagger category launches instead of putting the whole assortment live simultaneously. Give loyalty customers early access. Separate geographic markets. Move one category to a quieter fulfillment day. Limit quantities on inventory-constrained hero products. Reduce paid-media pressure once operational thresholds are reached.

The promotional calendar and fulfillment plan are really one operating decision.

If moving an event by 24 hours avoids excessive overtime, expedited transportation and thousands of delayed orders, that may protect more margin than squeezing every possible order into the original launch window.

Maximum demand is not always the objective.

Profitable, fulfillable demand is.

Measure the Promotion as an Inventory and Fulfillment Event, Not Just a Sales Event

The post-promotion review is where bad habits get reinforced.

Revenue beat plan. Conversion increased. Sell-through was strong.

Great. What happened to the inventory?

And what did it cost to fulfill those orders?

I would review a promotion across three areas.

First, the commercial and inventory outcome. Look at promotional uplift versus forecast, gross margin after discount, sell-through, stockout rate and post-promotion WOS. Look below category level too. A 70% category sell-through figure can hide a lot of ugly SKU and size-level positions.

Second, review fulfillment. Orders shipped within SLA, backlog by day, lines or orders processed per labor hour, split shipments, cancellations and carrier performance all matter.

Then look at incremental cost. Overtime, temporary labor, additional carrier collections, expedited transportation and any service recovery required after the event belong in the economics of the promotion. DHL notes that additional peak capacity and flexibility can come at a premium, which is exactly why sales volume alone is a poor measure of operational success.

Consider a footwear retailer running 25% off a category for a weekend.

The promotional report could look excellent. Strong revenue. High traffic. Great category sell-through.

Underneath that, three hero styles sell through core sizes on day one. Peripheral sizes remain heavily stocked. Saturday's order volume exceeds sustainable pick-pack capacity, creating a backlog into Sunday. Overtime is added, some orders ship late and the remaining assortment enters Monday with badly distorted size curves.

Was it a successful promotion?

Partly.

The demand was clearly there. But the business captured it inefficiently and left itself with a worse inventory mix. A good post-event review should make that visible, otherwise the same promotion gets repeated next season with the same problems.

Promotions are not just marketing events. They are inventory and fulfillment events.

The best promotion is not necessarily the one that creates maximum demand. It is the one that generates profitable demand the inventory position and fulfillment network can actually support.

That is a much tougher standard than revenue versus plan.

It is also a much more useful one.