Replenishment Strategies for Seasonal and Promotional Goods: Rhythm Beats Formulas

Why standard logic fails on these two categories

Standard replenishment assumes demand fluctuates around a stable baseline. Seasonal and promotional goods break exactly that assumption: their demand curves are manufactured, steep and unsustainable. Extrapolating with a baseline model means either under-stocking before the peak or — the single largest source of dead stock — continuing to order at peak inertia after it ends. For these goods, the core act shifts from "compute one quantity" to "ride a rhythm curve".

Seasonal goods: think in indices, not averages

Compute monthly (weekly before peak) seasonal indices: index(m) = average sales of month m across years ÷ overall monthly average. A sunscreen category might show 2.4 for June and 0.5 for September; next June's forecast = current baseline × 2.4. Use at least 2–3 years of history and recalibrate annually.

The rhythm has three phases:

1. Build-up (1–2 purchasing cycles before peak): front-load 60–70% of forecast peak volume. Stockout cost now exceeds overstock cost, so raise the service level.
2. Peak confirmation: use the first week or two of actual sell-through to correct remaining purchases — the most important human intervention of the year. The index is the prior; sell-through is the correction.
3. Wind-down: after mid-peak, switch the formula to stock-digestion mode; order only against certain gaps. Better to stock out slightly at season's end than carry inventory across seasons.

Iron rule: the last replenishment decision of a season matters more than the first. Carrying stock across a season usually costs more than two weeks of end-of-season stockouts.

Promotional goods: uplift thinking and a three-phase playbook

Uplift = promotional-period sales − baseline sales. Estimate the baseline by degrading data conditions: last comparable event's uplift coefficient × current baseline, adjusted for discount depth, placement and traffic forecasts; or curves from analogous products in the same price band; or, for brand-new items, a manual target calibrated by a small trial batch. Whatever the method, persist the coefficient and reconcile it against actuals after the event — that reconciliation is the only durable asset in promotional replenishment.

Then run three phases: build-up (one purchasing cycle out — order to the upper bound of the uplift forecast, bounded by warehouse capacity and capital); in-event (hourly/daily sell-through monitoring, threshold-triggered emergency orders or transfers, decisive sales throttling when lead times can't keep up); and wind-down (the week after — reset parameters to normal immediately, and evaluate cancelling or rescheduling unfulfilled in-transit orders).

Wind-down is the most neglected phase: weighted sales windows still carry peak values for one or two cycles, inflating suggestions. A parameter reset on the day the event ends should be part of the operating SOP, not left to memory. And event sales must be tagged and excluded — or modeled as a separate promotional series — before touching the baseline forecast.

The calendar is cross-departmental

Execution is half algorithm, half coordination: marketing calendars (event dates, placements, pricing) shared at least one purchasing cycle ahead; supplier capacity and lead-time elasticity confirmed for peak volumes; and a clearance plan (markdown ladder, bundling, channel diversion) written at stocking time, not improvised after the pile-up.

Seasonal and promotional replenishment tests rhythm management, not model precision: the courage to front-load, the speed of mid-peak correction, the decisiveness of the brake at the end. Decompose that curve into explicit phases with reviewable parameters and SOPs, and the business can capture peak upside without being drowned by the inventory left behind.