What Is Retail Allocation Software?
Retail allocation software is the system that decides how many units of each product, by size and pack, go to each store, channel or fulfilment node and when, distributing planned receipts and replenishment against store grades, size profiles, presentation minimums, capacity and current inventory so the plan's depth lands where it will sell.
Retail allocation software is the system that decides how many units of each product, by size and pack, go to each store, channel or fulfilment node and when, distributing planned receipts and replenishment against store grades, size profiles, presentation minimums, capacity and current inventory so the plan's depth lands where it will sell. It is the system of record for the placement decision: the grade and cluster each door belongs to, the size profile each cluster sells, the floor each door needs to present the product, the packs the goods ship in, and the on-hand, on-order and in-transit position every placement is calculated against. The merchandise plan tells it how much is landing and when; the assortment plan tells it which clusters and channels carry each option; the ERP tells it what is already in the building; the WMS takes what it decides and ships it.
This guide defines the software category and the function it serves. It is not a guide to allocating well, which inventory allocation and replenishment covers for apparel with a fully worked five-door example, and which allocation and replenishment by vertical covers across nine categories, with outdoor and sporting goods taken together. It answers two narrower questions that arrive together. First, what does "planning and allocation" mean as a function in a retail organisation, and how do planners, allocators and buyers divide the work? Second, what does a system have to do, and what does it have to hold, to be called retail allocation software rather than a distribution spreadsheet, a replenishment engine or an ERP's transfer screen? It covers the jobs the software does; the data model underneath them; how it connects to merchandise planning, assortment planning, the open-to-buy, the WMS, the OMS, the ERP and demand sensing; how it differs from replenishment software, allocation optimization and an ERP's distribution module; the methods it runs, explained by mechanism; what each of ten verticals needs it to carry natively; the criteria for choosing one; the ways implementations fail; and one worked example of the arithmetic the software exists to make cheap.
What "planning and allocation" means in a retail organisation
The phrase "planning and allocation" names both a process and the team that runs it, and a reader can arrive asking about either, so the function comes first. A retail business decides how much inventory it will own and where that inventory will sit, and those are two decisions made by two roles that share one object, the receipt.
Planners own the money and the flow. The merchandise planner builds the merchandise financial plan: sales, margin and inventory by department and month. From it comes the open-to-buy, the receipt budget by class and period, and from that the receipt plan, which says how many units of which classes land in which month. The planner's horizon is the season and the year, and the planner's grain is the department, the class and the month. The planner does not decide which door gets which units; the planner decides how many units there are to place, and when they arrive.
Allocators own the placement. The allocator takes a receipt the planner sized and the buyer committed and decides where it goes: which doors, which channels, which fulfilment nodes, how many units of each size, in what packs, on which day, and how much is held back at the DC for replenishment. After the initial placement the allocator owns the refill: the replenishment rules that keep each position stocked as it sells, the reallocation when a receipt slips or a door breaks, the transfer that moves units from a door that cannot finish them to one that can, and the consolidation that pulls a style back before its exit. The allocator's horizon is the receipt and the week; the allocator's grain is the SKU, the door and the day.
Buyers own the product. The buyer decides which styles, in which colors, from which vendor, at what cost, in what depth, and commits the purchase order. Where there is an assortment planner or merchandiser, that role sets the option count and the grid by cluster and channel before the buyer commits; the buyer turns the funded option list into orders. The buyer's grain is the style-color and the vendor.
| Role | Owns | Decides | Horizon | Grain | Hands to the next role |
|---|---|---|---|---|---|
| Merchandise planner | The financial plan, the open-to-buy, the receipt flow | How much is bought by class, and in which month it lands | Season and year | Department, class, month | A receipt plan with quantities and dates |
| Buyer, with the assortment planner | The product | Which options, from which vendor, at what cost, in what depth, for which clusters and channels | Season and delivery | Style-color, vendor | A committed purchase order with a grid by cluster and channel |
| Allocator | The placement and the refill | Which units, by size and pack, go to which door, channel or node, when, and how each position is refilled | Receipt and week | SKU, door, day | A distribution order to the DC; a replenishment order; a transfer |
The three hand off in that order, and the handoffs are where the function breaks when the systems that hold each role's work do not share a data model. The planner's receipt plan says 2,400 units of a class land in March; the buyer's purchase orders say 2,400 units across four styles with two ex-factory dates; the allocator's spreadsheet is built against the planned in-store date, and when one purchase order slips three weeks the allocation logic is placing units that are not in the building. The inventory and allocation teams page describes that week from the allocator's side; the point for this guide is that planning and allocation is one function precisely because the receipt is one object, and the software category exists to hold the allocator's half of it on the same object the planner's half was built on.
Why the two are named together rather than planning and buying, or buying and allocation, follows from the receipt flow. The planner and the allocator are the two roles whose work is measured in units by location by week rather than in styles: the planner sets how many units arrive in a month, the allocator sets where they sit in a week, and both are judged on the same measures, weeks of supply, in-stock rate, sell-through by door and the markdown at the end. In a small brand the two halves are one person; in a larger one they are a planning and allocation team with a planner per department and an allocator per department or region. Either way the function is the same: decide how much, decide where, and keep the two decisions consistent as the season moves. The rest of this guide is about the software that holds the second half.
What the software is for
The discipline is older than any software. A receipt lands at the DC, and somebody decides how to split it across the doors. In a brand without an allocation system that decision is made in a workbook: a tab per receipt, a row per door, a column per size, a grade column that drives a percentage, a presentation-minimum column that overrides it, and a pack-size cell that rounds the result. The workbook works until four things happen at once: the receipt arrives in two shipments instead of one, three doors have changed grade since the tab was built, the size curve the tab uses is the chain curve rather than the door's, and the replenishment for last month's receipt has to be calculated against an on-hand position that lives in the ERP and was exported on Monday.
Retail allocation software exists because allocation is a calculation that runs against a position that changes every day. The initial placement is a forecast made once per receipt, and it is wrong in detail at every door the moment the doors start selling. Everything after it, the refill, the reallocation, the transfer, the consolidation, is a recalculation against the current on-hand, on-order and in-transit position, by SKU, by location, on a cadence at least as fast as the fastest door can break. A workbook holds one state of that position; every cadence is a re-export and a rebuild. An allocation system holds the position as a live object, reads it from the ERP and the WMS on a schedule, and re-runs the rules against it, so that the question the allocator actually asks, "which doors need what this week, and do I have it?", gets a list instead of a morning.
Three consequences follow, and they organise the rest of this guide. First, the software is defined by the constraints it enforces and the position it reads, not by its screens: a tool that cannot hold a presentation minimum as a floor per door, a pack as the unit the goods ship in, and the on-order as part of the position is a distribution calculator, however good the grid looks. Second, it is a system of record for placement, not for the budget, the purchase order or the goods: it reads the receipt plan from merchandise planning, the position from the ERP, and hands a distribution order to the WMS, and each of those neighbours owns something it must not. Third, the allocation is the last planning decision before the goods move and the first one measured against actual sales, so the value of the software is mostly in how well it closes the loop: whether the sell-through it reads is read against the placement that produced it, so that the hindsight can tell a demand miss from a placement miss.
The jobs the software does
Eight jobs recur in every allocation system and in every allocation workbook that one replaces, because they are the decisions the function requires. Products package and name them differently; the decisions are stable. The table gives the one-line version and the subsections give what each job produces and what it hands to the next.
| Job | The question it answers | Works at the grain of | Hands on |
|---|---|---|---|
| Initial allocation of a receipt | How is this receipt split across doors, channels and the holdback on the day it lands? | Receipt by SKU by location | A distribution order to the DC and a holdback position |
| Store grading and clustering | Which doors behave alike, and how much does each sell of this kind of product? | Door by product group | A grade or cluster per door, with its demand index and size profile |
| Size-curve and pack-aware allocation | How is each door's quantity split by size, and rounded to the pack it ships in? | SKU by door | Whole units by size that sum to the door total, in shippable packs |
| Presentation minimums and capacity as constraints | What is the least a door needs to show the product, and the most it can hold? | Product group by door | A floor and a ceiling the allocation is clamped to |
| Replenishment to a target | How much does each position need back, given what it sold and what is already coming? | SKU by location by review | A replenishment order from the holdback or the vendor |
| Consolidation and inter-store transfers | Which doors cannot finish a style, and where should their units go? | SKU by door pair | A transfer order, or a pull-back to the DC |
| E-commerce and store fulfilment splits | How much of the receipt and the holdback is reserved for the online node, and which doors fulfil online orders? | SKU by node | A reserved quantity per node and a fulfilment eligibility by door |
| New-store and closing-store allocation | How is a door with no history opened, and how is one being closed run down? | Door by receipt | An opening allocation on an analogue door; a run-down schedule |
Initial allocation of a receipt
The first job is the one the category is named for: a receipt is at the DC, or is about to be committed at purchase order, and the system splits it. The split has two layers, how much goes to the doors and how much is held back, and then how the door quantity is divided across doors, channels and nodes. The methods section below explains the mechanisms; the point here is when the job runs. Allocating at purchase order, before the goods exist, lets the buy be cut to the door-level size and pack profile, so the vendor ships door-ready cartons; allocating at DC receipt, when the goods are counted, lets the split read the sales of the weeks since the order was placed. Software that supports only one of the two forces every receipt into the same timing, and the right timing differs by lead time and by how fast the doors become readable.
Store grading and clustering
Every allocation method reads a demand figure per door, and the grade or cluster is how that figure is built when the style itself has no history at the door, which is every new style. A grade ranks doors by how much they sell of a product group; a cluster groups doors that sell alike, by size profile, price sensitivity or climate as well as by volume. The system has to hold grades and clusters per product group, not per door, because a door that is grade A for outerwear can be grade C for swim, and it has to hold them with a date, because grades change and an allocation made on last year's grade is an allocation to a door that no longer exists. Store clustering for apparel covers how the groups are built and store clustering and localization by vertical covers the dimension that matters in each category; the software's job is to carry the result as a live attribute that every other job reads.
Size-curve and pack-aware allocation
A door quantity is not an allocation until it is split by size and rounded to the pack. The size curve is a profile of demand by size, and the system has to hold it by cluster or door, not as one chain curve, because the sizes a door sells decide whether the door finishes the style clean or breaks in week three with the core sizes gone and the tails left. The pack is the unit the goods ship in, a prepack of fixed ratio or a case of a single size, and rounding a door's need to packs changes the quantity and imposes the pack's size ratio on the door. The software's job is to do the arithmetic at every door, every receipt, with a rounding rule that keeps the sizes summing to the door total and shows the allocator what the rounding cost. Size and pack optimization covers the curve and the pack decision; case packs and planned depth covers what the pack does to depth; pre-pack vs singles is the glossary term.
Presentation minimums and fixture capacity as constraints
A presentation minimum is the least a door needs of an option for the option to be visible and sellable there: a full size run on a wall, a certain number of facings on a shelf, two of each shade on a gondola. Fixture capacity is the most a door can hold of a product group without the floor set breaking. The allocation is clamped between them, and the clamp has to be a field the system enforces, not a note the allocator remembers. A system without presentation minimums allocates small doors below the quantity that makes the option sell, and then reads the weak sales as weak demand. A system without capacity allocates more than the fixture holds, and the overflow lives in the stockroom until the markdown.
Replenishment to a target
Once the initial placement has sold for a week or two, the position at each door is a fact rather than a forecast, and the refill reads it. The target is stated as a model stock, a quantity per size per door, or as weeks of supply at the trailing rate, or as a minimum and maximum, and the system calculates the need per SKU per location at every review and fills it from the holdback or orders it from the vendor. The mechanism is in the methods section and in the replenishment trigger formula; the software's job is to count the right things: sellable stock only, on-order included so that a long lead time does not double-order, and a stop date so that the rule switches off before the exit. Core and NOOS replenishment programs covers the standing program for items that are never meant to sell through.
Consolidation and inter-store transfers
Some doors will not finish a style at full price, and the choice is to mark it down there, move it to a door that will finish it, or pull it back to the DC and ship it to the online node. The system's job is to find the pairs, a door with units it cannot sell in the weeks that remain and a door that would sell them, and to show the cost of the move against the markdown it avoids. A transfer that costs more in freight and handling than the margin it recovers is a markdown with extra steps, and the software has to show that arithmetic rather than generate every transfer the rule allows.
E-commerce and store fulfilment splits
The online node is a location that sells every size to every customer and has no fixture capacity, and it competes with the doors for the same receipt. The system has to hold the split between the node and the doors as a decision, reserve a quantity for the node from the receipt and the holdback, and, where stores fulfil online orders, hold which doors are eligible to fulfil which SKUs so that a door's last unit of a core size is not shipped to a customer three states away while a customer in the door is asking for it. Omnichannel assortment planning covers the planning side of that split; available-to-sell is the position the node reads.
New-store and closing-store allocation
A new door has no history, so its grade, size profile and demand index are borrowed from an analogue door or cluster and its opening allocation is built as a floor set rather than a fair share. A closing door is run down: replenishment stops, transfers out are prioritised and the last receipts skip it. Both are exceptions to every rule in the system, and the software has to hold them as attributes of the door with dates, so that every job above reads them. Planning for new store openings covers the opening side.
The data model it has to hold
Underneath the eight jobs is the set of objects that make them one system. A product can have good screens for every job above and still fail as allocation software if the objects below are missing, flattened or held as free text, because every job reads from them and every recalculation propagates through them. The table lists the objects; the paragraphs after it explain the three to check for first.
| Object | What it is | Why the allocation needs it |
|---|---|---|
| Merchandise hierarchy | Division to department to class to style to style-color to SKU | Grades, minimums and curves are held at a level of this tree and applied at SKU; the receipt plan arrives at class |
| Location hierarchy and attributes | Region to district to door, plus the DC and the online node, with each door's format, square footage, climate, opening date and closing date | Every placement is to a location; the attributes drive the grade, the capacity and the exceptions |
| Store grades and clusters | A rank or group per door per product group, with a demand index and an effective date | The demand figure every method reads when the style has no history at the door |
| Size profiles by cluster or door | Share of demand by size, per product group, per cluster or door | The door quantity is split by this; the chain curve is the wrong curve at every door |
| Pack configurations | The units and size ratio each product ships in, and whether loose units are allowed | The rounding is done to this, and the pack's ratio becomes the door's curve |
| Presentation minimums | The least of an option a door needs to present it, per product group per door or fixture type | The floor the allocation is clamped to |
| Fixture capacity | The most of a product group a door can hold | The ceiling the allocation is clamped to |
| Lead times | DC to door by lane; vendor to DC by vendor | The replenishment minimum is lead-time demand plus safety stock; the stop date is the exit minus the lead time |
| The inventory position | On hand, on order and in transit by SKU by location, read from the ERP and the WMS on a schedule | Every job after the initial placement is a calculation against this |
| The receipt plan | Quantities and dates by class and option, read from merchandise planning and the purchase orders | What there is to allocate, and when |
| Allocation and replenishment rules | Method, target, holdback share, rounding rule, review cadence, stop date, per product group | The logic itself, held as data so it can be changed without a rebuild |
| Exceptions | New doors, closing doors, remodels, events, embargoed SKUs, with dates | The overrides every job has to read |
| Actuals | Sales by SKU by door by day or week | The refill reads them; the hindsight reads them against the placement |
The three to check for first are the constraints: presentation minimums, fixture capacity and pack configurations, because a system can pass every screen-level test without enforcing any of them. A system without presentation minimums allocates a small door its fair share, which for a slow door and a shallow style is below the quantity that makes the style visible, and the hindsight then reads a presentation failure as a demand failure and grades the door down for next season. A system without capacity allocates to the demand index and discovers the fixture at floor set. A system without packs allocates in units and hands the DC a pick list it cannot pick, so the DC rounds to the carton on its own and the allocation of record is not what shipped.
The position deserves a sentence more. On order and in transit are part of the position, not a separate report. A replenishment rule that reads on hand alone fires again for units already on a truck, and on a long lead time it double-orders every review until the first order lands. The on-order glossary term covers the object; the point for the software category is that the position has to be read from the ERP and the WMS on a schedule the rule's cadence can rely on, and the schedule has to be visible, so the allocator knows whether the position on the screen is this morning's or Monday's. Data readiness for merchandise planning covers what has to be true of the hierarchy, the attributes and the actuals before any of the jobs above can run.
How it connects to its neighbours
Allocation software is defined as much by its seams as by its jobs, because it is the last planning system before the goods move and it reads from almost everything upstream of it. The table states what each neighbouring system owns and where the seam with allocation sits. Three of the seams, with merchandise planning, with the ERP and with the WMS, are the ones a workbook cannot hold as live relationships, and they are taken up after the table.
| Neighbour | What it owns | What it must not own | The seam with allocation software |
|---|---|---|---|
| Merchandise planning software | The financial plan, the open-to-buy, the receipt plan by class and month | Which door gets which units | The receipt plan is what there is to allocate, and when; the allocation's door-level cover rolls back up into the plan's inventory by month |
| Assortment planning software | Which options are carried, for which clusters and channels, how deep, by size and pack | The placement by door within the grid | The grid by cluster and channel is the allocation's target; the allocator distributes within it, door by door |
| Open-to-buy | The receipt budget by class, channel and period | Where received units sit | The holdback and the replenishment reserve are receipts the open-to-buy has to fund; a chase from the vendor spends open-to-buy |
| Buy planning | Quantities, costs, vendors, delivery dates and the purchase order | The split across doors | The purchase order is the receipt the allocation places; at purchase-order allocation the buy is cut to the door-level profile |
| Demand sensing software | The near-term read of how each SKU is selling at each location, days to a few weeks out | The placement decision | A demand signal the replenishment and reallocation jobs read as an input; the allocation decides what to do with it |
| ERP | Purchase orders, receipts, invoices, the stock ledger | Where units should go | Hands the allocation the on-hand, on-order and in-transit position; receives the allocation's transfers as stock movements |
| WMS | The physical inventory in the DC, the pick, pack and ship | Which doors need what | Receives a distribution order from the allocation and ships it; reports what actually left, which may differ from what was ordered |
| OMS | Customer orders and their routing to fulfilment nodes | The reserve per node | Reads the allocation's node reserves and door fulfilment eligibility; reports the demand the online node consumed |
| Supply chain planning software | The movement of goods from vendor to DC: lanes, lead times, capacity, inbound schedule | What the goods are for | Hands the allocation the in-DC date and the inbound slip; the allocation re-runs when the date moves |
| Space planning | Where merchandise physically goes in the door and how much room each category gets | What to send | Fixture capacity and presentation minimums, read into the allocation as the ceiling and the floor |
The seam with merchandise planning is the one that defines the function. The planner's receipt plan says how many units of a class land in a month; the allocator places them. In a workbook, the receipt plan is a figure typed into the allocation tab, and when the planner reforecasts the month, or the buyer's purchase orders come in over or under the plan, the tab is placing a quantity that no longer exists. Allocation software reads the receipt from the plan and the purchase order as a live object, and rolls the door-level result back up, so that the planner's inventory-by-month view shows what is actually in the doors, in the DC and on order, and the open-to-buy can see what the holdback and the replenishment reserve cost. Where the two systems share a data model the roll-up is the same object read at two grains; where they do not, it is an integration that has to be built and kept in step. What is merchandise planning software? describes the stack and its architectures; open-to-buy planning covers the budget the receipt plan is derived from.
The seam with the ERP is the position. Every job after the initial placement is a calculation against on hand, on order and in transit by SKU by location, and the ERP is where those facts live. The allocation system reads them on a schedule, and the two risks are the schedule being slower than the rule's cadence, so the rule fires on a stale position, and the position being read incompletely, so the rule fires on on-hand alone and ignores the truck. The seam runs the other way too: a transfer the allocation decides is a stock movement the ERP has to record, or the ledger and the placement disagree by the following morning.
The seam with the WMS is where the allocation becomes goods. The allocation produces a distribution order: so many units of each SKU to each door, in these packs, on this date. The WMS picks it, and what it picks may differ from what was ordered, because a carton was short, a SKU was damaged or a pack was substituted. The allocation of record has to be updated with what shipped, or the position the next review reads is wrong at every affected door. The inventory and allocation teams page draws the same boundary for RetailNorthstar: the planning and allocation-decision layer on one side, the WMS and TMS handling the physical inventory and transportation on the other.
The seam with demand sensing is the easiest to misdescribe, because a demand-sensing engine and a replenishment engine both produce a number per SKU per location per week. Demand sensing estimates how much will sell in the next days or weeks from high-frequency signals; allocation decides what to do about it, given the position, the holdback, the lead time and the exit. The sensed demand is an input to the replenishment and reallocation jobs, the replacement for the trailing rate at a demand inflection, and the decision is still the allocator's. What is demand sensing software? defines the neighbouring category; demand sensing in retail covers the discipline.
Allocation software vs replenishment software vs allocation optimization vs an ERP's distribution module vs spreadsheets
A search for retail allocation software returns five things that overlap it, and buyers meet all five in the same evaluation. The table draws the lines by mechanism. There are no vendor names here, because the boundaries are about what a tool does, not who sells it.
| Category | What it decides | Mechanism | Where it is strongest | Where it stops |
|---|---|---|---|---|
| Allocation software | How a receipt is split across doors, channels and the holdback, by size and pack, on the day it lands; and how positions are refilled, moved and consolidated afterwards | Forecast demand per door from grades and curves, clamped to minimums and capacity, rounded to packs; then need against the live position | Seasonal product with a first placement and a short window; a fleet with real differences between doors; size and pack complexity | It does not set the budget, commit the purchase order or move the goods |
| Replenishment software | How much each position gets back as it sells, from a holdback or a supplier | Need = target − (on hand + on order), on a trigger and a cadence, per SKU per location | Continuing items with history: core basics, consumables, a replenished shade ladder, hardgoods | The first placement of a new item, which has no position to read; the holdback decision; the exit |
| Allocation optimization | The placement that best meets an objective, such as expected sell-through or margin, across many constraints at once | A model that estimates demand per SKU per door and searches for the placement that maximises the objective subject to the constraints | Large fleets, many SKUs, many constraints, where the arithmetic of the baseline methods cannot be reasoned about door by door | Its estimates are as good as the history and the analogues; a new style in a new door is a modelled guess, and the model has to show its work |
| An ERP's distribution module | The transfer of recorded stock from one location to another | A transfer order against the stock ledger, by a fixed percentage or a manual list | Recording what moved, accurately, for the ledger | It has no grades, curves, minimums or forecast; it executes a split someone else decided |
| Spreadsheets | Whatever the builder wired in | Formulas against an exported position | A single receipt, a small fleet, a one-off decision | It holds one state of the position; every receipt and every review is a re-export and a rebuild, and nothing reads the result back |
Three of the boundaries deserve a paragraph each.
Allocation and replenishment are two mechanisms, and the category is at its best when one system holds both. Allocation places a receipt on a forecast, once, before the units have sold in the doors they are going to. Replenishment refills a position on a rule, repeatedly, in response to actual sales. The two meet at the holdback, the share of the receipt kept at the DC so that replenishment has units to work with, and the holdback is a decision only a system that sees both halves can make well: hold too little and the refill has nothing to send to the doors that broke; hold too much and the DC carries cover the doors should have been selling. A replenishment engine bought alone is excellent on the core and silent on the first placement of every new style; an allocation tool bought alone places the receipt and leaves the refill to a spreadsheet. Inventory allocation and replenishment works the two halves on one style, two weeks apart, and replenishment is the glossary term.
Allocation optimization is a method inside the category, not a separate category. Fair share, size-curve allocation and need-based replenishment are formulas an allocator can check by hand; an optimization model replaces the formulas with an estimate of demand per SKU per door and a search for the placement that maximises an objective under the constraints. Where the fleet is large and the constraints many, that is genuinely useful, because the formula methods cannot hold a dozen constraints at once and the allocator cannot reason about four hundred doors one at a time. Where the style is new, the door is new or the history was itself out of stock, the model's demand estimate is an analogue, and the allocator should be able to see the analogue and override it. Allocation optimization is the glossary term; the merchandise planning software guide's section on what "AI-assisted" honestly means applies here without change.
An ERP's distribution module records a movement; it does not decide one. The transfer screen in a stock system will move units from the DC to a door, and some will split a receipt by a fixed percentage per door. What it does not hold is the reason: the grade, the curve, the minimum, the capacity, the forecast, the holdback. A brand that allocates in its ERP is allocating in a spreadsheet whose formulas are fixed percentages, with the one advantage that the result is already in the ledger. The question that separates the two is the one in the selection table below: change a door's grade, and ask what else changed.
The methods, explained by mechanism
The allocation methods are formulas, and the formula pages carry each one with a calculator. This section states what each method does mechanically and what the software has to hold to run it, because a buyer evaluating allocation software is really evaluating whether the system can run these methods at the brand's own grain. The inventory allocation and replenishment guide states the same methods as formulas and resolves them in a five-door example; the worked example later in this guide resolves the first, second and fifth on a three-grade fleet.
Fair-share allocation gives each door a share of the units to allocate in proportion to its share of forecast demand: door units = units to allocate × (door demand ÷ total demand). The demand figure is a planned weekly rate or a season total, built from the door's grade and its history on comparable product. The mechanism is pure proportion, which is its strength and its limit: it puts units where the demand is and nothing else, so every other method is an adjustment to it. Run top-down, the same formula splits first to grades or clusters and then to the doors inside each, which gives the allocator a place to apply judgement where the history is deep before the arithmetic reaches the door where it is thin. The software has to hold a demand figure per door per product group, with the grade as its source when the style has no history. The fair-share allocation and door-level demand formulas carry the arithmetic.
Size-curve allocation splits a door's units by size: size units = door units × size share, with the shares summing to 100% and the units rounded to whole numbers that still sum to the door total. The rounding is where workbooks go wrong, because rounding each size to the nearest whole number can sum to one more or one less than the door total, and the standard fix is the largest-remainder rule: take the whole-number part of each size, and give the units that remain, one each, to the sizes with the largest fractional parts. The software has to hold a curve per cluster or door, not one chain curve, and apply the rule at every door on every receipt. The size curve allocation formula carries the arithmetic; the worked example below applies the rule to one door and shows nearest rounding failing.
Model-stock and min/max replenishment state a target position and refill to it. A model stock is a quantity per size per door, the position the door should hold on the floor; min/max states a minimum, lead-time demand plus safety stock, and a maximum, the minimum plus one order multiple, and orders up to the maximum when sellable on hand plus on order falls to or below the minimum. The mechanism is a comparison of the position against a target at every review, so the software has to hold the target per SKU per location, read the position including on order, and run the comparison on a cadence at least as fast as the fastest door can break. The replenishment trigger and safety stock formulas carry the two components.
Weeks-of-supply targets state the target as cover rather than as units: the door should hold so many weeks of its trailing rate, and need = (target weeks × trailing weekly rate) − on hand − on order, floored at zero. The mechanism scales the target with the rate, so a door that speeds up is topped higher without anyone changing its target, and a door that slows is topped lower. Its limit is the trailing rate itself, which lags a demand inflection: in the run-up to a peak the rate under-states next week, and after it the rate over-states. That is the seam where a sensed or forecast demand replaces the trailing rate, and the forward weeks of supply formula is the cover measure that pairs with it. The software has to hold the target as weeks per product group, compute the trailing rate over a window the allocator chose, and show the cover by door, because a chain-average cover hides every door that is about to break. The weeks of supply formula covers the measure.
Pack rounding and its cost is the method a demonstration can skip and the small doors cannot. Packs = door need ÷ pack size, rounded by the rule the business chose, and the door's units become packs × pack size, plus any loose units allowed. Rounding up protects the fast doors and spends the holdback faster; rounding down protects the holdback and leaves the slow doors below their minimum on the day the floor set opens. The ratio inside a prepack is also a size curve the door did not choose, so a pack-constrained program carries a loose-unit top-up for the sizes the pack under-supplies, or accepts that the small doors will carry the pack's tails. The software has to hold the pack configuration per SKU, apply the rounding rule per door, and show the allocator the difference between the need and the shipped quantity, because that difference is the cost of the pack and it is paid by the door that can least afford it. Case packs and planned depth covers the decision upstream of the allocation; the worked example below shows the arithmetic.
What the software has to handle natively, by vertical
The eight jobs are the same in every retail vertical. What differs is the object being allocated, what a complete position physically is, the lead time on the refill, and the force that ends the item's life, and a system that does not carry a vertical's grain natively pushes the allocator back into a spreadsheet beside it, which recreates the problem the system was bought to solve. Allocation and replenishment by vertical covers the discipline across these verticals, with outdoor and sporting goods taken together; the paragraphs below say what the software has to hold in each of the ten.
Apparel
The object is the style-color, allocated by size curve to a door, and the complete position is a size run with the core sizes in depth. The software has to hold a curve per cluster, apply the largest-remainder rule at every door, hold presentation minimums as a full run on the fixture, and run the refill from a DC holdback inside a window of weeks. Two things are specific to apparel drops. Fashion styles ship in deliveries that land every few weeks, so the initial-allocation job runs continuously rather than once a season, and the system has to allocate delivery three while delivery one is being replenished and delivery two is being consolidated. And size sparsity: a style bought shallow has one or two units of the tail sizes per door, so a single sale breaks the run, and the system has to decide at allocation whether the tails go to every door or are held at the DC and the online node for the doors that prove they sell them. The in-season chase is the reorder from the vendor that the holdback cannot cover.
Footwear
The object is the model in a colorway, allocated as a size run within a width, and a broken run is a stockout even when the door has pairs on the wall. The software has to hold the run as the unit of allocation and of presentation, so that a door receives complete runs or nothing, hold a size-and-width profile per door rather than per chain, and measure run completeness by door as its in-stock metric. Footwear carries a model year, so a carryover model is replenished across the changeover while the outgoing colorway is consolidated, and the system has to hold both states on one model. Where the brand sells to retail partners, the prebook sets the quantity before the season and the allocation is of the brand's own stock to its own doors and its online node around that commitment.
Accessories and bags
The object is the style in a colorway, often a hero color with a carryover core underneath it, and the allocation is as much about the core as the newness. The software has to run a standing replenishment on the evergreen core, allocated once and refilled by min/max indefinitely, while allocating the seasonal colorways as a drop. Attach rate matters here: an accessory sold alongside a parent category sells where the parent sells, so the door's demand index for the accessory borrows from the parent's, and the system has to be able to grade one product group by another's sales. Carryover means the exit is a decision rather than a date, and the consolidation job has to run on colorways rather than on the style.
Home and furniture
The object is the SKU in a finish, with long lead times and container minimums, and the allocation is of containers to a DC and of floor samples to doors. The software has to hold the container as the receipt and allocate a floor set of samples to each door against a planogram, hold the sellable inventory at the DC or a regional hub and fulfil customer orders from there, and run the refill by the container on an ocean lead time, which means the replenishment minimum is months of cover and is correct. Dealer prebooks in the wholesale channel commit quantities before the goods are made, so the allocation to dealers is a fulfilment of a commitment rather than a forecast. The floor set is the presentation minimum, and it is one unit of each sample, not a size run. Merchandise planning for home and furniture brands covers the planning side.
Outdoor
The object is a mixed line: a soft-goods style allocated by size curve beside a hard-goods SKU allocated by unit, across a model-year changeover. The software has to hold both grains in one allocation, hold the model year as a dimension so that carryover styles are replenished across the changeover while the outgoing year is consolidated and marked down, and run counter-seasonal categories on opposite calendars in the same fleet. Dealer prebooks commit a large share of the year's units before the season, so the brand's own allocation is of the remainder, the at-once stock and the online node's reserve, and the system has to show the prebook commitment against the inventory it is allocating from. Merchandise planning for outdoor brands covers the planning side.
Sporting goods
The object is the SKU by sport, and the calendar is the sport's season, which differs by sport and by region: the same fleet opens one category in a northern region while a southern region is in another. The software has to hold seasonality by sport and by region as attributes that drive the grade and the stop date, so that a door's demand index for a category is read against where that region is in that sport's season. Team and league prebooks commit quantities before the season and leave no refill lever, so the allocation of the prebook is a fulfilment and the allocation of the at-once stock is the forecast. Hardgoods carry long lives and are replenished by min/max; apparel and footwear in the same line carry the curve and the run. Merchandise planning for sporting goods brands covers the planning side.
Beauty
The object is the shade within a range, allocated as a set against a retail partner's planogram, and the complete position is the full shade range on the gondola. The software has to allocate the range as a unit, hold the planogram's facings as the presentation minimum per shade, run the refill on the core shades by min/max and on the seasonal shades as a drop, and hold the batch expiry and the retailer's required-remaining-life window as a stop on how much cover any position may carry; the period-after-opening on the pack governs the customer's use once it is opened, not how much unopened stock a door may hold. Gondola resets are scheduled events that change the facings, so the presentation minimum has a date, and the system has to hold the reset calendar and re-run the allocation to it. Testers and display units are not sellable stock and have to be excluded from the position the refill reads. Merchandise planning for health and beauty brands covers the planning side.
Toys and games
The object is the SKU inside a licensed window and a gifting peak, and the allocation is of case packs to facings. The software has to hold the licence window as a hard stop on replenishment, because stock that lands after the window closes cannot be sold, hold the gifting peak as the shape of the season so that the allocation front-loads the doors before the peak and the stop date falls well inside it, and allocate in case packs against a shelf plan where facings are the presentation minimum. Planning a licensed product window and planning the gifting calendar cover the two forces; merchandise planning for toy and game brands covers the planning side.
Baby and juvenile
The object is a hard-goods SKU with registry demand, and the allocation has to read a demand signal that arrives before the sale: a registry entry is a stated intention that converts over weeks, by location. The software has to read registry demand by door and node as an input to the grade and the refill, hold safety and compliance dates as a stop on how long a SKU may be sold, and run a standing replenishment on the core, because a registry item that is out of stock on the day the registrant's guest shops is a sale lost to another retailer. Planning with registry demand covers the signal; merchandise planning for baby and juvenile brands covers the planning side.
Jewelry and watches
The object is the individual piece, held one or two deep per door, on a precious-metal cost base that moves while the piece is on the floor, and the allocation is of pieces to cases. The software has to allocate at the unit grain, hold case capacity as the ceiling and one-of-each as the presentation minimum, run the refill as a one-for-one replacement on sale rather than as a curve, and carry the collection, and for watches the model year, so that a continuing piece is replenished across a collection change while the outgoing piece is consolidated into fewer doors. High unit value means a transfer between doors is worth its freight where an apparel transfer would not be, so the consolidation job runs more often and the cost arithmetic is different. Merchandise planning for jewelry and watch brands covers the planning side.
Selection criteria and the questions to ask
Four companion guides cover the evaluation of planning software in general: how to choose planning software, questions to ask planning software vendors, build vs. buy and how to evaluate the implementation model. This section narrows them to the criteria specific to the allocation layer, with the question that tests each in a demonstration. The test is always the same: bring your own receipt, your own fleet with its grades, and your own size curves and packs, and ask the vendor to allocate it in front of you.
| Criterion | What to look for | The question that tests it |
|---|---|---|
| The receipt seam | The receipt is read from the plan and the purchase order as a live object | "Slip this purchase order three weeks. What happens to the allocation, and where do I see it?" |
| Grades per product group | A grade or cluster per door per product group, with a date | "This door is grade A for outerwear and grade C for swim. Show me both, and change one" |
| Size curves by cluster | A curve per cluster or door, applied with a rounding rule that sums to the door total | "Allocate 72 units to this door on its own curve. Show me the rounding" |
| Presentation minimums as a constraint | A floor per product group per door the system enforces | "Give this slow door its fair share of a shallow style. What does the system do when it is below the minimum?" |
| Capacity as a constraint | A ceiling per product group per door | "Allocate more than this door's fixture holds. Is that visible before the DC picks it?" |
| Packs | Pack configurations per SKU, with the rounding rule and its cost shown | "This style ships in prepacks of twelve. Allocate the small doors and show me what the rounding cost" |
| The position | On hand, on order and in transit read on a stated schedule | "What is this door's position right now, when was it read, and does the refill count the truck?" |
| The holdback | A holdback share per receipt, held by size, released by rule | "Hold back a fifth of this receipt. What sizes are in the holdback after the door curves are applied?" |
| Replenishment targets | Model stock, min/max or weeks of supply per product group, with a stop date | "Set a four-week target on this style and a stop date at its exit. Show me the last order that can land and sell" |
| Transfers and consolidation | Door pairs with the cost of the move against the markdown avoided | "Find me the doors that will not finish this style, and show me whether moving the units beats marking them down" |
| The online node | A reserve per node and fulfilment eligibility by door | "Reserve units for the online node from this receipt, and stop this door fulfilling its last unit of a core size online" |
| Exceptions with dates | New doors, closing doors, remodels and events as attributes every job reads | "Open a door on an analogue next month and close another in eight weeks. Allocate this receipt" |
| Vertical grain | The object the brand actually allocates, carried natively | "Allocate a size run within a width", or "allocate a shade range to a planogram", or "allocate a container to a DC and samples to doors" |
| The WMS seam | What shipped is read back into the allocation of record | "The DC shipped two cartons short. What does the next review see?" |
| Optimization honesty | Clear answers on history, new doors and overrides | "What does your model do with a new style in a new door, and how do I see and override the analogue it used?" |
Two further criteria apply to any planning software and bear repeating here because the allocation layer makes them acute. Who changes a grade, a curve, a minimum or a pack configuration a year after go-live, and what it costs, decides whether the system keeps up with the fleet; the implementation model guide prices it. And how the position gets in, from which systems and how often, decides whether any job after the initial placement is possible at all; a replenishment rule on a position read weekly is a rule that is late by design.
Common failure modes
The failures below are the ones that recur in allocation implementations, stated with the mechanism that produces each and the fix. Several are failures of the discipline that the software makes visible rather than failures of the software, and that is the point: a system that holds the constraints as fields will show the allocator where the discipline is being skipped.
- Allocating history instead of plan. The initial allocation is built from last year's door-level sales of the comparable style, and last year's sales are read as last year's demand. But last year's sales at a door that broke in week three are not its demand; they are the sales the stockout allowed. The mechanism is that sales history is censored by availability, and the fix is to allocate against a planned rate by grade that the plan has set, using history as the index between doors rather than as the quantity, and to estimate lost sales at the doors that broke before grading them for next season. Lost sales is the glossary term.
- Pack rounding that over-stocks small doors. A slow door needs 24 units and ships in packs of twelve, so it receives two packs and is fine; a slower door needs 14 and receives two packs, 24 units, and carries nearly twice its demand because the rounding rule rounds up everywhere. The mechanism is that the pack is a fixed quantum and the cost of rounding falls on the doors whose need is smallest relative to it. The fix is a rounding rule that is allowed to round down where the presentation minimum permits, a loose-unit top-up on the sizes the pack under-supplies, and a pack size chosen for the small doors rather than the large ones, which is a buy decision that case packs and planned depth covers.
- Presentation minimums treated as demand. A door is lifted to its presentation minimum, and the lifted quantity is then read by the replenishment rule as the door's model stock, so the door is refilled to a quantity it was never expected to sell. The mechanism is that the minimum is a display requirement and the target is a demand figure, and a system that holds one field for both conflates them. The fix is two fields: a floor the initial allocation is clamped to, and a target the refill works to, with the refill allowed to let a slow door fall toward its floor as the exit approaches.
- Replenishing a fashion style as if it were NOOS. A seasonal style sells well in week two, the min/max rule fires, and units land in week eight of a twelve-week window with no time to sell at full price. The mechanism is a rule with no stop date applied to an item with a finite window. The fix is a stop date per style computed from the exit minus the lead time, after which the rule is switched off, and a separate standing program for the items that genuinely never exit, which core and NOOS replenishment programs and never out of stock cover.
- Allocation and plan disagreeing on the receipt month. The planner's receipt plan has the class landing in March, the purchase order has an ex-factory date that lands it in April, and the allocation was built against March. The doors are set for a receipt that is not in the building, the replenishment rule reads an on-order that is a month further away than it assumed, and the plan's inventory by month is wrong in both months. The mechanism is three systems holding the same date as three values, and the fix is one receipt object read by all three, re-run when the date moves.
- A chain size curve at every door. The buy was cut to the chain curve, and the allocation applies the same curve to every door, so a door whose customers skew small is broken in the small sizes by week three and carrying the large sizes to the markdown. The mechanism is averaging across doors that differ, and the fix is a curve per cluster held in the system and used at the buy as well as at the allocation, because replenishment cannot rebuild a curve that was wrong at the buy. How modern apparel brands approach sizing and replenishment establishes that limit.
- Reading the chain average. The weeks of supply for the style is fine at chain level while a third of the doors are broken and another third are carrying double cover. The mechanism is that allocation failures are door-and-size failures, and every chain-level measure averages them away. The fix is to measure where the decision is made: in-stock, fill rate, weeks of supply and sell-through by door and by size, with the chain figure as a roll-up rather than the view.
A worked example: one receipt, three grades, a size curve, a minimum and a pack
The arithmetic below is the first, third and fourth of the eight jobs, done by hand for one style-color on one receipt; the fifth, replenishment to a target, is worked two weeks after the placement in the sibling inventory allocation and replenishment guide. Every figure is illustrative, chosen so the division comes out cleanly; none is a benchmark. The point is not the numbers but the shape of the calculation, the way each constraint moves units, and the fact that an allocation system has to redo all of it every time a grade, a curve, a pack or the receipt itself changes.
The inputs:
- A receipt of 2,400 units of one style-color of a women's knit top lands at the DC.
- The fleet is 30 doors in three grades for this product group: 6 grade A doors planned at 12 units a week each, 10 grade B doors planned at 6 a week, and 14 grade C doors planned at 2 a week.
- The brand holds back 20% of the receipt at the DC for replenishment, so 480 units are held and 1,920 go to the doors at the initial allocation.
- The presentation minimum for this style is a full six-size run with depth in the core sizes: XS 2, S 5, M 8, L 8, XL 5, XXL 2, which is 30 units per door.
- Grade A and B doors receive open stock. Grade C doors are shipped in prepacks of 12 in the ratio XS 1, S 2, M 3, L 3, XL 2, XXL 1, with no loose units, and the rounding rule is to round up where rounding down would break the presentation minimum.
- The grade B size curve for this product group is XS 5%, S 18%, M 29%, L 26%, XL 15%, XXL 7%.
Step 1: fair share by grade, then door. The planned weekly rates sum to 6 × 12 + 10 × 6 + 14 × 2 = 72 + 60 + 28 = 160 units a week across the fleet. Dividing 1,920 by 160 gives 12, so the initial allocation is 12 weeks of cover at the planned rate at every door. Grade A receives 1,920 × (72 ÷ 160) = 864 units, or 144 per door; grade B receives 1,920 × (60 ÷ 160) = 720 units, or 72 per door; grade C receives 1,920 × (28 ÷ 160) = 336 units, or 24 per door. The three grade totals sum to 864 + 720 + 336 = 1,920.
Step 2: the presentation minimum. A grade C door's fair share is 24 units, below the 30-unit minimum, so each grade C door is lifted to 30. The lift is 6 units per door across 14 doors, 84 units, and it has to be funded from somewhere; the brand funds it from the holdback rather than by scaling the A and B doors below their demand, so the holdback falls from 480 to 396, and the refill has 84 fewer units to work with. The sibling guide makes the opposite choice, taking the lift from the doors above the minimum and leaving the holdback intact; either is defensible, and the system's job is to show what the choice cost. Grade C now holds 14 × 30 = 420 units. The grade C doors are carrying 15 weeks of cover at their planned rate against 12 at the other grades, and that difference is the cost of being present in the door: the minimum is a display requirement, and the system has to hold it as a floor rather than let the refill read it as the door's demand.
Step 3: the size split for one grade B door. A grade B door receives 72 units, split on the grade B curve with the largest-remainder rule.
| Size | Share | 72 × share | Whole part | Fraction | Final units |
|---|---|---|---|---|---|
| XS | 5% | 3.60 | 3 | 0.60 | 3 |
| S | 18% | 12.96 | 12 | 0.96 | 13 |
| M | 29% | 20.88 | 20 | 0.88 | 21 |
| L | 26% | 18.72 | 18 | 0.72 | 19 |
| XL | 15% | 10.80 | 10 | 0.80 | 11 |
| XXL | 7% | 5.04 | 5 | 0.04 | 5 |
| Total | 100% | 72.00 | 68 | 72 |
The whole parts sum to 68, so four units remain, and they go one each to the four sizes with the largest fractions: S at 0.96, M at 0.88, XL at 0.80 and L at 0.72. The door receives XS 3, S 13, M 21, L 19, XL 11, XXL 5, which sums to 72. Rounding each size to the nearest whole number instead would have given XS 4, S 13, M 21, L 19, XL 11, XXL 5, which sums to 73, one unit the door does not have; across ten grade B doors the error would be ten units the DC cannot pick. By grade, the ten doors receive XS 30, S 130, M 210, L 190, XL 110 and XXL 50, which sums to 720.
Step 4: pack rounding at the grade C doors. A grade C door's quantity after the minimum is 30 units, and it ships in prepacks of 12. Dividing 30 by 12 gives 2.5 packs. Two packs is 24 units, below the 30-unit minimum, so the rule rounds up to three packs, or 36 units per door. The pack ratio of 1-2-3-3-2-1 gives each door XS 3, S 6, M 9, L 9, XL 6, XXL 3, which sums to 36 and happens to meet or exceed the minimum in every size. The rounding costs another 6 units per door, 84 units across the 14 doors, again funded from the holdback, which falls from 396 to 312. Grade C now holds 14 × 36 = 504 units, and each grade C door is carrying 18 weeks of cover at its planned rate against the 12 the allocation intended: 3 weeks of the excess is the minimum and 3 weeks is the pack, and both are paid by the doors least able to sell them.
The result, and the check.
| Line | Fair share | After the presentation minimum | After pack rounding |
|---|---|---|---|
| Grade A, 6 doors at 144 | 864 | 864 | 864 |
| Grade B, 10 doors at 72 | 720 | 720 | 720 |
| Grade C, 14 doors | 336 (24 each) | 420 (30 each) | 504 (36 each) |
| Units in the doors | 1,920 | 2,004 | 2,088 |
| DC holdback | 480 | 396 | 312 |
| Receipt | 2,400 | 2,400 | 2,400 |
At every stage the doors plus the holdback sum to the receipt: 1,920 + 480, 2,004 + 396 and 2,088 + 312 are each 2,400, and the final placement is 864 + 720 + 504 + 312 = 2,400. The holdback that started at 20% of the receipt is 312 ÷ 2,400 = 13% after the two constraints have taken their 168 units, and it is the residual by size after every door's curve and pack have been applied, so it is not the chain curve either: it is short in whichever sizes the door curves and the pack ratio drew most heavily, which is the sizes the fast doors will chase first. The inventory allocation and replenishment guide works that holdback-by-size effect and the replenishment two weeks later on a different fleet.
The what-if that shows why the system has to hold the relationships: suppose the pack for the grade C doors is changed at the buy to a 6-unit prepack in the ratio XS 0, S 1, M 2, L 2, XL 1, XXL 0. Five packs, 30 units, meets the door total exactly and the pack rounding costs nothing, but the pack carries no XS and no XXL, so the door is below its presentation minimum in the two tail sizes on the day it opens, and the 84 units the pack saved are now a loose-unit top-up of 2 XS and 2 XXL per door, 56 units, that the DC has to pick as singles. One change to the pack configuration moved the grade C quantity, the holdback, the size position in every grade C door and the DC's pick list. In a workbook that is a rebuild of every tab the pack touches; in an allocation system it is the same receipt with one input changed, which is the whole case for the category.
Where RetailNorthstar fits
RetailNorthstar's allocation planning page describes the capability in these terms: distribute inventory across channels and doors to protect full-price sell-through, with pre-season and in-season allocation in one workflow. It lists channel allocation, in which confirmed buy quantities are allocated across DTC, wholesale and store channels with the splits connected to the buy plan rather than managed in a separate file; AI allocation recommendations by door, generated from historical sell-through patterns, store capacity and style-level performance data; door-level distribution, tiering stores by category sell-through rather than overall revenue; pre-season allocation planning tied directly to confirmed buy quantities and OTB; in-season reallocation, identifying doors over-stocked and under-stocked against their plans with reallocation recommendations surfaced; and a buy plan to allocation handoff in which confirmed buy quantities flow into the allocation workflow with no re-entry. Recommendations are a starting point, the page says: planners review, adjust by exception and confirm the allocation plan before it is finalized. The inventory and allocation teams page draws the system boundary: RetailNorthstar is the planning and allocation-decision layer, the WMS handles physical inventory and the TMS handles transportation, with receipts, on-hand by location and shipment status flowing in and allocation plans flowing out to the WMS for execution.
The evaluation criteria in this guide apply to RetailNorthstar as they apply to any product in the category, and a brand should run them against it: bring the receipt, the fleet with its grades, the curves and the packs, and allocate it in the demonstration. Which of the eight jobs a brand needs, at what grain, and with which neighbours already in place, are questions this guide is meant to help settle before any demonstration, this one included.
Related resources
- What Is Markdown Optimization Software? — the software category: the decisions it owns, how the price-response model and the optimizer work, and a worked example of two markdown paths
- Inventory Allocation and Replenishment and Allocation and Replenishment by Vertical — the discipline the software serves, in the flagship category and across nine
- What Is Merchandise Planning Software? and What Is Assortment Planning Software? — the layers above this one
- What Is Demand Sensing Software? and Supply Chain Software vs Merchandise Planning Software — the neighbours on either side
- Store Clustering for Apparel and Store Clustering and Localization by Vertical — the grades and clusters every method reads
- Size and Pack Optimization and Case Packs and Planned Depth — the curve and the pack
- Core and NOOS Replenishment Programs and Planning for New Store Openings — the standing program and the exception
- Questions to Ask Planning Software Vendors — the evaluation
- Fair-Share Allocation, Size Curve Allocation, Replenishment Trigger, Weeks of Supply and Fill Rate — the formulas
- Allocation Optimization, Replenishment, Store Clustering, Size Curve, Pre-Pack vs Singles, Safety Stock, Never Out of Stock, In-Stock Rate and Fill Rate — the terms
- Allocation Planning and For Inventory and Allocation Teams — the product and the team page
See how RetailNorthstar builds pre-season allocation from the confirmed buy and OTB, with allocation recommendations by door from historical sell-through, store capacity and style-level performance.
Start Free Trial →No credit card. No commitments.
Common questions
What is retail allocation software?
Retail allocation software is the system that decides how many units of each product, by size and pack, go to each store, channel or fulfilment node and when, distributing planned receipts and replenishment against store grades, size profiles, presentation minimums, capacity and current inventory so the plan's depth lands where it will sell. It is the system of record for the placement decision: the grade and cluster each door belongs to, the size profile each cluster sells, the floor each door needs to present the product, the packs the goods ship in, and the on-hand, on-order and in-transit position every placement is calculated against.
What does planning and allocation mean in a retail organisation?
Planning and allocation names the two halves of the function that decides how much inventory a business owns and where it sits. Planners own the financial plan, the open-to-buy and the receipt flow: how much is bought by class, and in which month it lands. Allocators own the distribution of that inventory to locations: which doors, channels and fulfilment nodes receive which units, by size, on which receipt, and how each position is refilled as it sells. Buyers own the product itself: which styles, from which vendors, at what cost and in what depth. The three roles meet at the receipt: the planner sizes it, the buyer commits it and the allocator places it, and the function is named for the two roles that share the receipt flow from the plan to the door.
What is the difference between allocation software and replenishment software?
Allocation software places a receipt on a forecast, once per receipt, before the units have sold in the doors they are going to; replenishment software refills a position on a rule, on a cadence, in response to actual sales. Allocation answers how a receipt is divided across doors and channels on the day it lands; replenishment answers how much each position gets back as it sells, from a DC holdback or from a supplier. The two meet at the holdback, the share of a receipt kept at the DC so that replenishment has units to work with, and a system that only does one of them leaves the other to a spreadsheet beside it.
What data does retail allocation software need to hold?
The merchandise hierarchy from division to SKU; the location hierarchy with each door's attributes, grade and cluster; a size profile per cluster or door; the pack configurations the goods ship in; a presentation minimum per product group per door; fixture capacity per door; lead times from DC to door and from vendor to DC; the current on-hand, on-order and in-transit position by SKU by location; the receipt plan with its quantities and dates; and the allocation and replenishment rules themselves, with their targets and exceptions. The three to check for first are the constraints: presentation minimums, capacity and pack configurations, because a system can pass every screen-level test without enforcing any of them.
How does allocation software connect to merchandise planning, the ERP and the WMS?
Merchandise planning sits above it and hands it the receipt plan and the assortment grid: how many units of each option land in which month, for which clusters and channels. The ERP holds the purchase orders, receipts and stock ledger, and hands the allocation system the on-hand, on-order and in-transit position it calculates against. The WMS executes the physical pick, pack and ship at the DC, and the OMS routes customer orders to fulfilment nodes. Allocation software sits between them: it reads the plan and the position, decides the placement, and hands a distribution order to the WMS; it does not set the budget, place the purchase order or move the goods.
What is fair-share allocation?
Fair-share allocation gives each door a share of the units to allocate in proportion to its share of forecast demand: door units equal units to allocate multiplied by door demand divided by total demand. In this guide's worked example, 1,920 of a 2,400-unit receipt go to 30 doors whose planned weekly rates sum to 160, so the allocation is 12 weeks of cover at the planned rate: 144 units to each grade A door planned at 12 a week, 72 to each grade B door planned at 6, and 24 to each grade C door planned at 2. Fair share is the baseline; presentation minimums, size curves and pack rounding are applied on top of it.
How do presentation minimums and pack rounding change an allocation?
They lift small doors above their fair share, and the lift has to be funded from somewhere. In the worked example, fair share gives a grade C door 24 units, but the door's presentation minimum is 30, so the door is lifted to 30 and 14 grade C doors take 84 extra units from the DC holdback. The grade C doors ship in 12-unit packs, so 30 units becomes 2.5 packs, which rounds up to 3 packs of 12, or 36 units per door, taking another 84 units. The doors end up holding 2,088 units against a fair-share figure of 1,920, and the holdback falls from 480 to 312 units; 864 to grade A, 720 to grade B, 504 to grade C and 312 at the DC sum to the 2,400-unit receipt.
Does the same retail allocation software work for footwear, beauty, furniture and other verticals?
The jobs are the same in every vertical; the object being allocated is not. Footwear allocates a size run within a width to a door and treats a broken run as a stockout. Home and furniture allocate containers to a DC and floor samples to doors, on an ocean lead time, against a floor set. Beauty allocates shade ranges as a set and cannot hold cover past the batch expiry and the retailer's required-remaining-life window. Sporting goods and outdoor allocate against a dealer prebook and a model year. Toys allocate into a licensed window and a gifting peak. Jewelry allocates individual pieces on a precious-metal cost base. A system that does not carry the brand's own allocation grain natively pushes the allocator back into a spreadsheet beside it, so the evaluation should allocate the brand's own object before any other feature is looked at.
Share this guide with your team
Copy a link or a pre-written message for Slack, Teams, or email.
// Know where your operation stands
Apply this to your planning operation.
The free Apparel Planning Maturity Assessment benchmarks your operation and tells you exactly which gaps to fix first.