Allocation and Replenishment by Vertical

What allocation and replenishment are

Allocation is the decision that turns a bought quantity into a position: which units go to which door, channel, wholesale account or dealer, at what grain, and when. Replenishment is the rule that refills that position as it sells — from a holdback the brand kept in its own warehouse, or from a supplier — on a trigger set before the first unit shipped. Both sit downstream of the buy and upstream of the markdown, which makes them the two places where a correct buy is converted into sell-through or quietly wasted. The right model for both is vertical-specific, because the three things a model is built from — the grain the product sells at, the lead time on the refill, and whatever forces the item to leave the assortment — differ in each category, and a rule tuned for one fires on the wrong object in the next.

This guide extends two apparel pieces rather than repeating them: allocation and replenishment for apparel covers initial allocation, holdback and the trigger in the flagship category, and sizing and replenishment in apparel establishes that replenishment cannot rebuild a wrong size curve. It is the third piece in the by-vertical series, after the merchandise hierarchy and the planning calendar; the exit side is covered separately.

The invariant: three questions, nine answers

Every allocation model answers the same three questions. What is the object being allocated — the unit whose absence costs a sale? Where does a refill come from, and how long does it take? And what ends the item's life, so replenishment knows when to stop? The table is the guide in one view; the sections that follow are the reasoning behind each row.

Apparel: size-curve allocation, door clusters, packs and the chase

The apparel grain is the style-color by size, allocated to a door or a cluster, and the first decision is whose size curve a door receives. The brand-average curve sent everywhere produces residual in both directions — broken cores in doors that skew one way, fringe residual in doors that skew the other — so the curve is set at the grain the evidence supports: store clustering carries a cluster curve where door history is thin and a door curve where it is deep, the door-level demand index sets depth within the cluster, and size-curve allocation resolves depth into units by size.

The second decision is pack versus singles. A pre-pack is cheap to pick and imposes its curve on every door that receives one; singles cost more and let the curve flex per door. The workable split is packs for the initial floor set and singles for replenishment, because at floor-set depth the pack's rounding error is small against the door total, while each replenishment unit is a correction to one size in one door that a pack cannot make. The third decision, the holdback, is taken in the receipt plan before the first receipt lands, and it is shared: DTC and wholesale allocations pull from the same holdback, so a wholesale reorder filled from it is a DTC replenishment that no longer exists. Carryover and newness run on different rules inside the same style-color matrix — carryover styles sit on the replenishment rule because their curve is proven, while newness is allocated once per delivery window and read before it earns a trigger at all. The fourth decision is what makes apparel unusual: the in-season chase. Apparel is the one category where a supplier reorder is routinely open for part of the season, so replenishment has two sources with two lead times — holdback in days, chase in weeks — and the item carries one minimum per source, switching as the chase window closes.

Footwear: size runs, widths, breakage and pairs per door

Footwear allocates a run, not a count. The sellable object is a model-color in a size run complete enough that a customer who wants the model can find their size; a door with the middle of the curve gone is not a low-stock door but a broken run and a dead wall slot. Allocation starts from the run minimum per door — the sizes that make the model presentable and sellable there — and adds core-size depth on top. Widths multiply the run: a model in two widths is two runs on the wall and two in the stockroom, and the second width rarely earns full-fleet distribution, so it is rationed to the doors whose fit history supports it, with a transfer path for the rest. Assortment planning for footwear covers the buy side.

Replenishment is breakage-driven. The trigger is not total pairs on hand but the loss of a size the run cannot do without, and the order is a run repair — the specific sizes that restore the run — rather than a top-up to a pair count, which fires late by construction because the run collapses while the count still looks healthy. Pairs per door is the depth target above the trigger: the run minimum plus enough core depth to survive one replenishment cycle. Because the model year owns the exit, replenishment on a changing-over model stops on a date, and the final repairs are sized so the outgoing chassis clears before the incoming one takes the wall — the sequencing in planning a model-year changeover.

Accessories & bags: evergreen core, hero color and attach rate

Accessories carry two clocks in one assortment. The evergreen core — black leather, standard hardware, stable velocity — has no season: it is placed once in every door that carries the category and then run on a replenishment rule, refilled from the supplier on a lead time set by leather and hardware. It is the closest thing in fashion to a classic min/max item, and the worked example below is built on it. The seasonal colorways are the opposite — allocated once against the collection, not replenished, exited on the collection's end date. Planning accessories lines covers the buy that sets those two pools up.

Within the seasonal layer, the hero color goes deep into the doors with proven color demand and shallow everywhere else; the secondary colors carry the depth risk, bought thinner and placed in fewer doors, because a fashion color that residualizes cannot be carried into a later season the way a core can. The third mechanism is attach rate: the share of transactions in a host category that include the accessory. An accessory that attaches — a belt to denim, a bag to outerwear — is placed in proportion to the host's allocation rather than to its own history, because its demand is generated by the host's traffic. That puts the belt in the door that got the denim depth, a decision the accessory's own history would never suggest, since that history only records where it was placed last time.

Home & furniture: containers, ocean lead times, WOS, landed cost and floor samples

In home and furniture the container is the order unit and the allocation unit. Allocation is not units to doors; it is containers to distribution centers, and floor samples to doors. The floor sample is a display asset, not inventory available to sell, and a system that counts it as on-hand overstates cover in every door and under-fires every trigger; floor samples are allocated once per introduction, and replacing them when they age is a separate decision from replenishment.

Ocean lead times set the min/max. The minimum is lead-time demand plus a safety buffer — large, because the lead time runs to months and transit variability is the dominant risk — and the maximum is the minimum plus a container, because a partial container is not an economic order. The result is a weeks of supply target that would read as gross overstock in apparel and is simply correct here. Cover is not free, and the number it is priced at is landed cost — first cost plus freight, duty, insurance and inbound handling — not the factory FOB; OTB planning for home goods covers how that cover reconciles with the open-to-buy. Dropship removes replenishment for part of the assortment: items allocated to no door and fulfilled to order from the supplier have no min/max at the brand, and a supplier availability check replaces the rule. The wider treatment is in merchandise planning for home and furniture brands.

Outdoor & sporting goods: model years, prebooks versus at-once, counter-seasonal categories and hard-goods min/max

Outdoor and sporting goods lines hold two replenishment worlds side by side. On the prebook side, dealer prebooks allocate the model year's production before the season opens: the allocation is the prebook itself, the factory builds against it, and once capacity is allocated there is no refill lever. What remains is the at-once pool — whatever the brand held back for dealers who under-booked — so the real replenishment decision on prebook product is the preseason holdback, sized before any read exists. Model-year transitions add a stop: allocation of the outgoing model ends at the changeover, and the at-once pool is run down deliberately so the two models do not compete for the same dealer floor. Counter-seasonal categories sharpen the point — the refill on a snow category cannot be placed in the season it sells, and an early warm season shortens the window before the read is complete — and MAP pricing removes price as the rebalancing lever, leaving reallocation between dealers and channels as the in-season tool that moves units.

The other world is hard goods and consumables — wax, balls, grips, tubes, small parts — which run on a classic min/max with a domestic lead time, the one part of the line where a weekly trigger works exactly as it does in a warehouse. The mistake is one rule for the whole line: min/max on prebook product reserves open-to-buy for a reorder that cannot be placed, and prebook logic on consumables lets a fast-moving accessory stock out between seasonal reviews. The category guides for outdoor and sporting goods cover the buy side.

Health & beauty: shelf life, shade ladders, launch fill versus replenishment, testers and damages

Beauty replenishment is bounded by a date before it is bounded by demand. Every batch carries a shelf life or a period-after-opening rule, and a retailer will refuse receipt inside a required-remaining-life window, so the maximum in min/max is capped by the date rather than by the order multiple — cover that cannot ship before the ceiling is not cover. When dating rules cap your weeks of supply computes the ceiling; the allocation consequence is that a fast-selling shade and a slow-selling shade from the same batch have the same last ship date, and the slow one is the one the rule has to watch.

The shade ladder is the allocation object. A range has to stay complete in a door, because a missing shade is a stockout for every customer of that shade and an intact display for everyone else, so the trigger runs per shade with ladder completeness as the constraint, and the ends of the ladder are where it breaks. Launch fill and replenishment are two pools with two rules: the launch is allocated once to fill the planogram on the reset date, read for a launch period, and only then handed to the replenishment-heavy core rule with its own stock pool and cadence. Testers and damages sit inside the allocation, not beside it — a tester is a non-sellable unit per door per shade and a damage allowance is a predictable loss on fragile packaging, and both are funded up front or the sellable count is overstated from day one. See merchandise planning for health and beauty brands.

Toys & games: Q4 concentration, retailer resets and licensed windows

Toys concentrate the year's selling into one peak, which collapses replenishment into a preseason decision plus a domestic scramble. Allocation is planogram-driven: facings per store on the retailer's reset date set the initial quantity, in case and inner-pack multiples rather than units — the arithmetic in case packs and planned depth. Inside the peak, replenishment can only come from what is already in a domestic DC, because a factory reorder placed once the peak is readable lands after it; the replenishment decision is therefore the size of the domestic holdback committed before the reset.

Retailer resets set both the intake and the exit: an item leaves the shelf on the retailer's schedule regardless of sell-through, so the final shipment is sized to sell down by the reset, not to the velocity, or it becomes the retailer's markdown and the brand's returns. Licensed windows add a harder stop — replenishment ends on the contract date whatever the item is doing, and the last order is placed at the window end minus the lead time and sized to the window end. See planning a licensed product window and merchandise planning for toy and game brands.

Baby & juvenile hard goods: registry-driven long-tail replenishment and recall traceability

Juvenile hard goods run on a long tail with a forward read. A registry creates demand before the purchase — the item is added weeks or months ahead of the buy — and the demand persists, because the registry buyer arrives on their own date and the item has to be there when they do. Replenishment is long-tail by design: low weekly velocity, long product life, and a stockout that costs more than the sale, since a registry buyer who finds the item unavailable moves the whole registry. The registry count is a replenishment input, not just a forecast input; planning with registry demand covers how it is read.

Traceability changes what the allocation record has to hold. A recall is a lot-level event, and the business has to find which doors and customers received which production lot, so allocation and replenishment records carry the lot rather than treating units as fungible across runs. Certification changes end an item's life the way a model-year changeover does elsewhere. See merchandise planning for baby and juvenile brands.

Jewelry & watches: high unit value, low depth, memo and security

Jewelry allocates the piece, not the curve. Unit value is high, depth per door is one or two units per style, and the sale of a piece is the replenishment trigger — a one-for-one refill from bench or made-to-order capacity — which makes the min/max almost trivial and the question of which pieces sit in which doors almost everything. Carat and metal price exposure is why that one-for-one refill is priced at the time of refill rather than at the original buy — the replacement piece carries the metal price of the day it is made, not the day the sold piece was bought. Ring sizing is the exception where a run reappears, handled by sizing after sale rather than by stocking the run. The low-markdown culture of the category also changes what a slow piece means: it is not a markdown candidate but a rotation candidate, moved to a door where it may sell before capital, not price, decides its exit.

Memo and consignment change what allocation means: goods on memo remain the brand's inventory, the allocation is a placement, the replenishment is a rotation of pieces between doors as each proves or fails to sell, and the record has to show where every serialized piece is. Security is the allocation constraint with no analogue elsewhere — what a door can hold is bounded by what it can secure and insure, so the ceiling on depth per door is the safe and the insurance limit rather than the fixture. Gifting peaks raise depth temporarily, and the gifting calendar decides where; the capital side of holding pieces that never date is in merchandise planning for jewelry and watch brands.

The allocation grain decision

The allocation grain is the smallest object the allocation decides and the largest object it executes at. The rule that generates the nine answers above is short: allocate at the grain the door sells at, and no finer than the history can support. The first half catches the coarse error — footwear allocated as a pair count breaks the run while the count looks healthy, beauty allocated as a range total breaks the ladder at its ends, furniture allocated in units fails to reconcile with the container it ships in. The second half catches the fine error — apparel allocated door-by-size where door history is a handful of units per size turns noise into decisions, and the corrections chase each other, the argument the merchandise hierarchy guide makes for planning levels. Clustering is the standard answer to the fine error — decide the curve at the cluster, execute it at the door — and it applies wherever the sellable object has a curve, which is apparel and footwear, and not where it does not.

Min/max and safety stock by lead-time profile

The min/max rule is the same in every category. The minimum is the demand expected during the refill lead time plus a safety buffer; the maximum is the minimum plus one order quantity, rounded to whatever the supplier ships in — a pack, a carton, a run, a container. The site's own formulas are safety stock and the replenishment trigger; what changes between categories is their inputs.

Four lead-time profiles cover the nine categories. A refill measured in days — the warehouse holdback in apparel and footwear, the domestic DC in toys inside the peak — needs a small buffer, because a miss is corrected next cycle. A refill measured in weeks — a domestic accessories supplier, sporting goods consumables, an apparel chase against held fabric — needs a buffer sized to demand variability over those weeks. A refill measured in months — a furniture factory slot plus ocean transit, a beauty batch fill, juvenile hard goods on a certified configuration — needs a buffer sized mostly to lead-time variability, because a short slip on a months-long lead time loses more cover than any plausible demand swing. And a refill that does not exist — the dealer prebook, the licensed window, the outgoing model year — has no min/max at all; the decision it replaces is how much to hold back before the season, made once. Two caps sit above the rule: beauty's maximum is bounded by the dating ceiling regardless of the order multiple, and jewelry's by security and capital regardless of demand.

Illustrative example: the same rule in two verticals

The figures below are illustrative, chosen because they divide cleanly. They are not benchmarks, not targets, and not drawn from any brand. Both items use the site's own formulas: safety stock is average weekly demand times safety days divided by seven; the minimum is lead time in weeks times average weekly demand, plus safety stock; the maximum is the minimum plus one order multiple.

Item A is an evergreen accessories core — a black leather belt refilled from a domestic supplier. Average weekly demand is 30 units, the lead time is 3 weeks, and the safety buffer is 7 days because the lead time is short and stable. Safety stock is 30 × 7 ÷ 7 = 30 units. The minimum is (3 × 30) + 30 = 120 units. The supplier ships in cartons of 60, so the maximum is 120 + 60 = 180 units. At the minimum the belt holds 120 ÷ 30 = 4 weeks of supply; at the maximum, 180 ÷ 30 = 6 weeks.

Item B is a home and furniture core — a dining chair in one finish, refilled by the container from an overseas factory. Average weekly demand is 10 units, a third of the belt's; the lead time is 14 weeks covering the factory slot and ocean transit; the safety buffer is 28 days because a slipped vessel is the dominant risk. Safety stock is 10 × 28 ÷ 7 = 40 units. The minimum is (14 × 10) + 40 = 180 units. The container holds 120 chairs, so the maximum is 180 + 120 = 300 units. At the minimum the chair holds 180 ÷ 10 = 18 weeks of supply; at the maximum, 300 ÷ 10 = 30 weeks.

Same formula, same label, and the chair carries five times the belt's cover at maximum while selling a third as fast. Neither item is overstocked and neither is understocked; the cover follows the lead time, and the lead time follows the category. A weeks-of-supply target borrowed from the accessories business would fire the chair's trigger every week and never fill it, because the container would be judged an overstock the day it landed. The difference is not free — the chair's maximum is priced at landed cost and tied up for months — which is why cover in long-lead categories is a capital decision, not a stock decision.

The replenishment trigger

The trigger fires when sellable on-hand plus on-order falls to or below the minimum, and each term is a decision. Sellable excludes the furniture floor sample, the beauty tester, the jewelry piece on memo and the damaged carton, or the rule reads a position that cannot be sold and fires late. On-order includes what is already in transit, or a long-lead category double-orders every time the trigger is checked before the first order lands. The minimum is per item and per location, and where an item has two refill sources it carries one minimum per source. Cadence follows the shortest lead time in the rule: a weekly review is right for a days-profile refill and pointless for a months-profile one, where the trigger is checked against the vessel schedule instead.

Every trigger also has a stop date, computed rather than chosen. The last replenishment that can land and sell is placed at the exit date minus the lead time, and after that date the rule is switched off — otherwise it keeps filling a position the calendar is about to close, and the residual it builds is invisible until the reset, the changeover or the dating ceiling arrives. The exit dates are the subject of the planning calendar; clearing the residual is the subject of markdown and exit strategy by vertical. Wholesale adds one more constraint: the trigger is only as good as the fill rate behind it, because an order the DC cannot ship complete restores a broken run in the plan and not in the door.

How allocation feeds the hindsight

The season hindsight has to separate a demand miss from a placement miss before it writes a buy rule, and only the allocation record can do that. A style that stocked out may have been under-bought; or bought correctly and sent to the wrong doors; or sent to the right doors on the wrong curve; or held back in the warehouse while the doors broke; or replenished on a trigger that read floor samples as sellable. Sales history alone reads every one of those as an underbuy and writes a deeper buy rule for next season, which funds a placement failure with inventory — the execution-error trap the hindsight guide describes, arriving through the allocation door.

So the record that has to survive the season is the allocation decision and the curve it used, the holdback and when it was released, the replenishment log with each trigger's source and lead time, the run repairs in footwear, the ladder breaks in beauty, the lot in juvenile hard goods, the memo rotations in jewelry. Held where the next allocation is made, that record lets the hindsight write a curve correction, a holdback change or a trigger reset instead of a deeper buy; held in a spreadsheet rebuilt each season from sales, the placement story is gone. Holding allocation, sell-through and the trigger in one connected workflow is the practical answer, and it is the approach behind allocation planning.

Related resources

• →Allocation & Replenishment for Apparel — the flagship-category playbook this guide extends
• →Sizing and Replenishment in Apparel — why replenishment cannot rebuild a wrong size curve
• →The Planning Calendar by Vertical — the exit dates that set every trigger's stop date
• →Markdown and Exit Strategy by Vertical — what happens to the residual the trigger did not prevent
• →Safety Stock Formula — the buffer term in the minimum
• →Replenishment Trigger Formula — the minimum itself, with the calculator
• →Fair-Share Allocation Formula — proportional allocation across doors, and when to override it
• →How to Hindsight a Season — where the allocation record is finally read