Skip to main content
50 min readassortment planning softwareassortment planning tool

What Is Assortment Planning Software?

Assortment planning software is the system a brand or retailer uses to decide which options to carry in a season, for which channels and store clusters, and how deep to buy each one by size and pack, inside the open-to-buy that funds them.

Assortment planning software is the system a brand or retailer uses to decide which options to carry in a season, for which channels and store clusters, and how deep to buy each one by size and pack, inside the open-to-buy that funds them, holding that plan as one connected object from the first placeholder to the committed buy. It is the system of record for the assortment plan: the option count by class and price tier, the depth behind each option, the grid that says which cluster and channel carries it, and the size and pack profile the buy will be cut to. The open-to-buy tells it how much money there is; the line plan tells it what could exist; the buy plan takes what it decides and turns it into purchase orders.

This guide defines the software category. It is not a guide to the discipline, which what is assortment planning? covers for apparel and assortment planning by vertical covers across ten categories, and it is not a vendor comparison, which best assortment planning tools is. It answers a narrower question: what does a system have to do, and what does it have to hold, to be called assortment planning software rather than a planning spreadsheet, a line-sheet tool or a reporting layer? It covers the jobs the software does; the data model underneath them; how it connects to merchandise financial planning, open-to-buy, line planning, allocation, PLM and ERP; how it differs from merchandise planning software, assortment optimization and category management tools; what assortment optimization honestly means and where it stops working; the criteria for choosing one; what each of ten verticals needs it to carry natively; the ways implementations fail; and one worked example of the arithmetic the software exists to make cheap. It is the assortment sibling of what is merchandise planning software?, which defines the wider category this one sits inside.

What the software is for

The discipline comes first, and the discipline is older than any software. Assortment planning is the decision about which options a brand carries in a season, channel or location, and how deep each is bought, within the budget the open-to-buy allows. In a brand without a planning system that decision still gets made; it gets made in an assortment workbook, with a tab per cluster or channel, a column per option, a row per door, and a size curve pasted in from last season. The workbook works until three things happen at once: the open-to-buy moves after the grid is built, the line changes after the grid is built, and the grid has to be re-cut for a second channel.

Assortment planning software exists because the assortment is a plan that changes shape while it is being made. The option count depends on the depth, the depth depends on the presentation minimum and the door count, the door count depends on the cluster definition, and the whole thing depends on an open-to-buy figure that finance is still revising. A workbook holds one state of that system at a time; every change is a rebuild. A planning system holds the relationships, so a change to any one input re-derives the others, and the question a merchandiser actually asks, "if I add this option, what do I give up?", gets a number instead of an afternoon.

Three consequences follow, and they organise the rest of this guide. First, the software is defined by the objects it holds and the constraints it enforces, not by its screens: a tool that cannot hold a presentation minimum as a floor on depth per door is a grid editor, however good the grid looks. Second, it is a system of record for an intended assortment, not for transactions or for the product's specification; it reads the product record and the actuals from elsewhere and hands a buy downstream. Third, the assortment plan is one layer of a stack, below the merchandise financial plan and the open-to-buy, beside the line plan, above the buy and the allocation, and the value of the software is mostly in how well it holds its seams with those layers. What is merchandise planning software? describes the whole stack; this guide describes the layer where options are chosen.

The jobs the software does

Seven jobs recur in every assortment planning system and in every assortment workbook that one replaces, because they are the decisions the discipline 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 actually produces and what it hands to the next.

JobThe question it answersWorks at the grain ofHands on
Option count and depth against the OTB envelopeHow many options can this budget fund at a depth that will sell?Class, price tier, cluster, deliveryA funded option count and planned depth per option
Assortment grids by cluster and channelWhich options go to which clusters and channels, and at what depth there?Option by cluster or channelA grid, the matrix the allocation will read
Attribute and good-better-best balanceIs the mix right across silhouette, fabrication, color, price tier, core and new?Attribute roll-ups of the option listAttribute targets and the variance to them
Size and pack profilesHow is each option's depth split by size, and how is it packed?SKU, size within optionA size profile and pack configuration per option per cluster
Placeholder-to-style evolutionWhat is in each slot of the plan today, and what has changed since the plan was built?Placeholder, then style, then style-color, then SKUA line that keeps its financial frame as it becomes real product
What-if on breadth versus depthWhat does one more option cost, and what does one fewer buy?The plan as a wholeCompared scenarios with their financial consequences
In-season reassortment and chaseWhat is selling, what is not, and what do we add, cut, move or reorder this week?Option by door by weekReassortment decisions, chase orders, transfers, exits

Option count and depth against the OTB envelope

The first job is the arithmetic that the worked example at the end of this guide walks through: take the receipt budget the open-to-buy allows for a class, a price tier, a cluster and a delivery, convert it to units at the planned ticket price, and decide how many options it funds at what depth. The two are a trade: more options means shallower depth, and depth has a floor, set by the presentation minimum per door, the size curve and the vendor's order minimums, below which an option is present but not sellable. The software's job is to keep the option count and the depth in one view with the budget they share, so that adding an option shows what it does to the depth of the others, and cutting one shows what it frees. What an option costs sets out the costs of an option that do not scale with its depth; breadth vs depth is the trade in one glossary entry.

Assortment grids by cluster and channel

The second job is placement before allocation. Not every option goes everywhere: a cluster of small doors cannot present the full range, a wholesale account books from a range that differs from the DTC assortment, a marketplace carries the core and not the fashion. The assortment grid is the matrix of options by cluster or channel, each cell holding whether the option is carried there and at what planned depth. It is the object the allocator will later read as a target, and the object the wholesale team will read as the range. The software has to hold the grid natively, with store clusters and channels as planning dimensions, roll the cells up to the option's total depth and to the cluster's total option count, and show when a cluster has more options than its fixtures can carry. Localized assortment planning and omnichannel assortment planning cover the two axes of the grid; store clustering for apparel covers how the clusters are built.

Attribute and good-better-best balance

An assortment is a list of options, but it is judged as a mix. The merchandiser needs to see the list rolled up by silhouette, fabrication, color family, fit, price tier, core against seasonal, carryover against new, and to compare each roll-up with a target set before the options were chosen. Attribute balance is the job that makes the assortment a plan rather than a shopping list: a target share of the class in the better tier, or a maximum share in a single silhouette, is set first, and the options are chosen to hit it. Good-better-best is the price-architecture form of the same job, and it has a specific failure the software should make visible: tiers that differ in price but not in what the customer can see of the product cannibalise each other, so the roll-up by tier needs to sit next to the attributes that justify the tier. The software has to carry attributes as planning dimensions, not as text columns, so that the roll-ups are live as options are added and cut. Product mix optimization for apparel covers the mix decisions.

Size and pack profiles

Depth per option is a total; the buy is cut by size, and in many categories it is packed. The size profile splits an option's depth across sizes from a size curve built at the class or style level and adjusted for the cluster, and the pack profile decides whether the units ship as prepacks in a fixed ratio or as open stock picked by size. The job sits in assortment planning rather than in allocation because the size and pack profile changes what the depth means: an option with a depth of 480 units in prepacks of twelve is forty packs, and a cluster of forty doors gets one each, while the same 480 units as open stock can be curved differently for a door that sells large. The software has to hold a size profile per option per cluster, round it to the pack configuration, show where the rounding pushes the buy over or under the plan, and pass the profile intact to the buy and the allocation. Size and pack optimization covers the method; case packs and planned depth covers what the pack does to the depth arithmetic; pre-pack vs singles is the glossary term.

Placeholder-to-style evolution

An assortment plan is built before the product exists. In the first pass the options are placeholders: a slot in the class at a price tier with a planned depth and a planned attribute set, "better-tier knit top, core color, delivery two", without a style number, a sample or a cost. As the line develops, placeholders become styles, styles become style-colors, and style-colors become SKUs with sizes, costs and vendors. The software's job is to keep the financial frame attached to the slot as the slot fills, so that the plan built on placeholders in month one is the same plan, with the same option count and depth, that is reviewed on real product in month four, and the variance between the two is visible rather than lost in a rebuild. A system that only holds real styles forces the plan to start late, after the line is drawn; a system that only holds placeholders cannot carry the plan into the buy. The ladder from placeholder to SKU is the spine of the data model, taken up below, and the handoff from line to assortment is the subject of line planning vs assortment planning.

What-if on breadth versus depth

The sixth job is comparison. Every one of the decisions above has an alternative, and the merchandiser's question is rarely "what is the plan?" but "what is the plan if?". If the cluster drops four doors. If the hero style is bought a third deeper and the seventh option is cut. If the better tier takes a larger share. If the delivery is split in two. In a workbook each of these is a copy of the file, and the copies drift apart. In a planning system the scenarios are versions of one plan, held side by side, each showing its option count, its depth per option, its spend against the envelope and its attribute roll-ups. Scenario comparison is possible without any model at all; what the shared data model adds is that the scenarios are cheap, which in a workbook they are not, and cheap scenarios get run. Scenario planning for retail covers the practice.

In-season reassortment and chase

The seventh job is the assortment plan done again with actuals in it. Once the season opens, sell-through by option by door by week comes back, and the questions change: which options to chase, which to cut from the clusters where they are not selling, which to move between doors, which to exit early, and whether the next delivery's grid should change given what the first delivery did. Reassortment is the in-season version of the grid job, and chase is the in-season version of the depth job. The system that holds the pre-season plan has to be the system that reads the in-season actuals against it, at the grain the decisions were made, or the reassortment happens in a different file from the plan and the two never meet again. The in-season chase covers the chase decision; how to hindsight a season covers the read that closes the loop into the next season's plan.

The data model it has to hold

Underneath the seven jobs is the set of objects that make them one system. A product can have excellent screens for every job above and still fail as assortment planning software if the objects below are missing, flattened or held as free text, because every job reads from them and every change propagates through them. The table lists the objects; the paragraphs after it explain the three constraints to check for first.

ObjectWhat it isWhy the assortment plan needs it
Merchandise hierarchyDivision to department to class to style to style-color to SKUEvery option count, depth and roll-up is at a level of this tree; the OTB envelope arrives at class, the buy leaves at SKU
Product attributesSilhouette, fabrication, color family, fit, end use, material, finish, shade, licence, and whatever else the vertical plans onThe attribute balance job rolls the option list up by these; they have to be dimensions, not text
Store clusters and gradesGroups of doors that share a demand profile, a size profile and a fixture capacity; a grade is the cluster's capacity tierThe grid is built by cluster; depth per option per cluster depends on the cluster's door count and minimums
ChannelsDTC, stores, wholesale accounts, marketplaces, outlet, each with its own assortment and targetsThe grid's second axis; the same option is carried at different depth, or not at all, by channel
Price tiersGood, better, best, or the vertical's own architecture, with target price points and target margin per tierThe envelope can be split by tier before options are chosen; the tier roll-up is the price architecture check
The placeholder to SKU ladderA slot that becomes a style, a style-color and a sized SKU as the line developsThe plan has to be built before the product exists and carried into the buy after it does
Presentation minimumsThe smallest quantity and range of an option a door needs for the option to sell thereA floor on depth per option per door; without it the option count is unconstrained
Fixture capacityHow many options and units a door, or a cluster's doors, can physically present per classA ceiling on option count per cluster; without it the grid overfills the floor
Calendar and deliveriesRetail weeks, seasons, delivery drops or flows, the model year where the vertical uses itOptions are planned into deliveries; depth is planned against weeks on the floor
Open-to-buy envelopeThe receipt budget by class, tier, channel and period, read from the OTBThe constraint every option count and depth is tested against
Size curves and pack configurationsSize profile by class and cluster; prepack ratios and case quantitiesThe size and pack job reads them; the buy is cut by them
ActualsSales, stock, receipts and on-order by SKU by door by week, read from the ERP and channelsThe in-season job and the hindsight both read them against the plan

The three to check for first are the constraints: presentation minimums, fixture capacity and the placeholder ladder, because a system can pass every screen-level test without them. A system without presentation minimums lets the option count float free of what a door can sell, and the plan that results looks fully funded and is not, because the depth per door has quietly fallen below the quantity that makes the option visible. A system without fixture capacity lets the grid put more options into a cluster than its doors can present, and the overflow is discovered by the store team at floor set. A system without a placeholder ladder either forces planning to start after the line is drawn, which is too late for the financial frame to shape the line, or cannot carry the plan into the buy without re-entry. The assortment planning by vertical guide calls the first of these minimum viable presentation and treats it as the floor on depth per door in every category; the point for the software category is that a constraint has to be a field the system enforces, not a note the planner remembers.

The hierarchy and the attributes deserve a sentence more. The merchandise hierarchy is the tree every module of a planning system shares, and the assortment plan is the layer that moves down it furthest before the buy, from the class the envelope arrives at to the style-color the option is committed at. Attributes are not part of the tree; they cut across it, so that a silhouette roll-up can span several classes and a color-family roll-up can span the whole season. A system that holds attributes as columns on a product record but cannot roll the plan up by them has the data and not the job. Data readiness for merchandise planning covers what has to be true of the hierarchy and the attributes before any of the jobs above can run.

How it connects to its neighbours

Assortment planning software is defined as much by its seams as by its jobs, because every decision it makes is constrained from above, informed from the side and consumed from below. The table states what each neighbouring system owns and where the seam with assortment planning sits. Two of these seams, with the open-to-buy and with the line plan, are the two a workbook cannot hold as live relationships, because it carries the envelope as a typed cell and the line as a re-keyed sheet, and they are taken up after the table.

NeighbourWhat it ownsWhat it must not ownThe seam with assortment planning
Merchandise financial planningSales, margin and inventory targets by department and periodWhich options are carriedSets the frame the open-to-buy is derived from; the assortment's tier and class mix has to roll up into it
Open-to-buyThe receipt budget by class, channel and periodThe option count, which it constrains but does not chooseThe envelope: the assortment spends it, option by option, and every depth change moves the position
Line planning and visual line boardsWhat could exist: the styles, colorways, fabrications and stories design proposes for the seasonWhich of them are bought, and how deepThe line is the menu; the assortment is the order; placeholders in the assortment become styles from the line
Buy planningQuantities, costs, vendors, delivery dates and the commitment to the factoryThe option mix, which it executesThe assortment hands a funded option list with depth, size and pack profiles; the buy rounds it to minimums and commits it
Allocation and replenishmentWhich doors receive which units, pre-season and as receipts land; how continuing items are refilledThe grid, which it reads as a targetThe assortment grid by cluster is the allocation's target; the allocator distributes within it
PLMHow the product is made: tech pack, bill of materials, samples, cost versionsHow many to buy, where to placeThe product record and attributes the assortment plans on; the costing target the assortment's margin assumes
ERPOrders, receipts, invoices, stock on hand, the ledgerWhat to carry nextThe buy goes out as purchase orders; sales, stock and receipts come back as the actuals the in-season job reads
Demand forecasting and demand sensingHow much of an existing item will sell, by location and week, and how fast that read updatesThe season's option mixA forecast baseline for continuing options and a replenishment driver for the core; an input, not the decision
Space planningWhere merchandise physically goes and how much room each category getsWhat to buy and how deepFixture capacity, read into the assortment as a ceiling on options per cluster

The seam with the open-to-buy is the one that defines the category. In a workbook the open-to-buy is a figure typed into a cell at the top of the assortment sheet, and the sheet is built to it. When finance revises the envelope, or when a buy committed in another category relieves the shared position, the cell is stale, and the grid built under it is funded against a budget that has moved. Assortment planning software reads the envelope from the open-to-buy as a live constraint, shows the plan's spend against it as options are added and cut, and pushes the committed depth back as spend. What is open-to-buy planning? covers the OTB; MFP, assortment plan and open-to-buy covers how the three layers stack; the OTB planning page describes the open-to-buy as a live workflow.

The seam with the line plan is where the plan gets its options. Line planning decides what could exist, in design's vocabulary of stories, silhouettes, fabrications and colorways; assortment planning decides what is bought, in the merchandiser's vocabulary of option counts, tiers, clusters and depth. In a brand without a system the handoff is a re-keying: someone types the line sheet into the assortment workbook, and from that moment the two drift. Where the line is planned visually, on boards, the seam is cleaner when the board and the plan are the same object, so that a style placed on a board is an option in the plan with a slot, a tier and a depth. Line planning vs assortment planning draws the boundary; the visual line and visual assortment planning pages describe the board-as-plan approach; line planning is the glossary term.

The seam with demand planning is the easiest to misdescribe, because the two categories share the word "demand" and little else. Demand forecasting and demand sensing read how an existing item is selling and project it forward; they are strongest on continuing items with history and weakest on a new option that has none. Assortment planning uses that read as an input, as the baseline depth for a carryover option or the replenishment driver for a never-out-of-stock core, and makes the decision the forecast cannot: whether the option is carried at all. Demand planning vs merchandise planning separates the disciplines, and what is demand sensing software? defines the neighbouring software category. The seams below the assortment, with the buy and with allocation, are described on the buy planning and allocation pages and in supply chain software vs merchandise planning software, which maps which system owns each decision from the line plan to the receipt.

Assortment planning software vs merchandise planning software vs assortment optimization vs category management tools

A search for assortment planning software returns four categories that overlap it, and buyers meet all four in the same evaluation. The table draws the lines. Each boundary has a companion guide that argues it in full, and this guide does not re-argue them.

CategoryWhat it decidesGrainWhere it is strongestWhere it stopsCompanion guide
Assortment planning softwareWhich options to carry, for which clusters and channels, how deep, by size and pack, inside the OTBOption by cluster by deliverySeasonal product, fashion newness, the breadth-depth trade, placeholder planningIt does not set the financial frame or post the transactionThis guide
Merchandise planning softwareThe whole stack: financial plan, OTB, assortment, buy, size and pack, allocation, in-season reforecastDepartment and class down to SKUConnecting the layers on one data modelDepth in any one layer is traded for breadth across themWhat is merchandise planning software?
Assortment optimizationWhich options from a candidate set best meet an objective, using demand transfer and like-item models on historyItem by store, on continuing items with historyReplenished categories with long histories; rationalising a range; localising a gridFashion newness without history; the financial frame; the line itselfThe next section of this guide
Category management toolsHow big each category should be, what role it plays, how its space and budget are set, across a retailer's floorCategory by storeRetailer-side range reviews, planogram-driven categories, supplier-led category plansThe option-level plan inside the category; the brand's lineCategory management vs assortment management

Two of the boundaries deserve a paragraph each.

Assortment planning software is one layer of merchandise planning software, and a brand can buy the layer or the stack. A point solution for assortment planning does the seven jobs above in depth and integrates with whatever holds the open-to-buy and the buy; a connected platform holds the assortment as one module beside them. The trade is depth against reconciliation: a point solution can model the grid, the attributes or the placeholder ladder more finely, because that layer is all it has to model, and leaves the seam with the open-to-buy and the buy to an integration that has to be built and maintained. The architecture section of what is merchandise planning software? covers suite modules, point solutions and connected platforms; the point here is that a brand evaluating assortment planning software is choosing a layer, and should know whether it is buying the layer alone.

Category management tools come from the retailer side and assortment planning software from the brand side, and they meet at the shelf. Category management, as the discipline grew up in grocery and mass retail, decides the role, the size, the space and the budget of a category across a retailer's floor, often in collaboration with the category's leading supplier; its tools are built around planograms, space-to-sales and range reviews. Assortment planning decides the options inside the category. A brand that sells through retail partners meets category management as the process that decides whether its range is listed and how many facings it gets; a retailer runs both, category management above assortment planning. Category planning is the glossary term for the retailer-side layer, and SKU rationalization is the decision where the two meet.

What "assortment optimization" means in practice

Assortment optimization is the capability in this category that most needs defining before it is evaluated, so it is worth saying plainly what the models do, where the mechanism holds and where it does not. There are no benchmark figures in this section, because the honest answer to "how much does it improve the assortment?" depends on the history the model has to work with, and that is the point.

Demand transfer is the estimate of what happens to the demand for an option when it is removed: how much of it moves to the options that remain, how much leaves the category and how much leaves the store. It is the model behind every rationalisation decision, because cutting an option is only a saving if its demand transfers. The mechanism needs two things: a history of the option and its neighbours selling side by side, and a measure of their similarity, by attribute or by observed substitution. Where both exist, in a replenished category with a stable range, the estimate is useful and the range review it supports is better than one made on rate of sale alone. SKU rationalization is the decision it serves.

Like-item modelling is the analogue method for an option with no history of its own: find the past options most like it by attribute, price tier and placement, and use their performance as the baseline for its depth and its grid. This is the model that lets an optimization engine say anything at all about a new option, and its quality is bounded by how well the attributes describe what the customer is actually responding to. For a new colorway of a continuing style the analogue is close and the estimate is good; for a new silhouette in a new fabrication at a new price point the analogue is a class average, and the estimate is a starting point for a product judgement rather than a substitute for one.

Cannibalisation estimates are demand transfer run in the other direction: what adding an option takes from the options already in the plan. They are the model behind the breadth-versus-depth trade when it is made analytically rather than by rule, because the true cost of a seventh option is not only the depth it takes from the other six but the demand it takes from them. The same conditions apply: the estimate is as good as the history and the similarity measure.

Where the mechanism holds, assortment optimization is genuinely useful: continuing items, replenished categories, carryover programs, a core range being rationalised, a grid being localised across clusters on the basis of how the same items have sold in each. Those conditions hold wherever the range is replenished rather than bought once: a beauty franchise's shade ladder, a home brand's stocked finishes, a footwear brand's carryover models, a sporting goods brand's hardgoods, an apparel brand's never-out-of-stock program.

Where it breaks is fashion newness without history. A seasonal line that is largely new by option count, which an illustrative split might put at six new options in every ten, gives the model six options with no history and four with some, and the analogues for the six are only as good as the attributes. The model will produce a number for every option, and the numbers for the new ones will be confident, because the model does not know what it does not know. The failure is not that optimization is used but that its output is read as a measurement rather than as a baseline. A buyer evaluating the capability should ask three things: what history the model was built on, what it does with an option that has none, and how the planner overrides it and sees that the override has been applied. The questions to ask vendors guide's AI questions cover the first and the third; the second is the one to add.

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 that are specific to the assortment layer, with the question that tests each in a demonstration. The test is always the same: bring your own category, your own cluster definition and your own open-to-buy figure, and ask the vendor to build the plan in front of you.

CriterionWhat to look forThe question that tests it
The OTB seamThe envelope is read from the open-to-buy as a live constraint and spend is pushed back"Revise the envelope by ten percent while the grid is open. What changes, and where do I see it?"
Placeholder planningThe plan can be built on slots before styles exist and carried into the buy after they do"Build a class on placeholders, then replace three with real styles. Does the frame survive?"
Presentation minimums as a constraintA floor on depth per option per door that the system enforces, by cluster"Add an option the envelope cannot fund at the minimum. What does the system do?"
Fixture capacity as a constraintA ceiling on options per cluster per class"Put more options into a small-door cluster than its fixtures hold. Is that visible before floor set?"
Attribute roll-upsAttributes as planning dimensions with targets, live as options change"Show me the better-tier share and the new-versus-carryover share as I cut an option"
The gridOptions by cluster and channel, with depth per cell, rolling up both ways"Carry this option in two of four clusters and in wholesale only. What does allocation receive?"
Size and pack profilesA size profile per option per cluster, rounded to the pack, with the rounding visible"Change this option from open stock to prepacks of twelve. What happens to the depth and the buy?"
Scenario comparisonVersions of one plan held side by side with their consequences"Hold the current plan and a version with one fewer option and a deeper hero. Compare them on one screen"
In-season reassortmentThe same plan reads the actuals and supports chase, cut, move and exit"It is week four. Show me which options to chase in which clusters, from this plan"
Vertical grainThe option grain the brand actually plans at, carried natively"Plan pairs by size within width", or "plan a shade ladder as a set", or "plan a frame in four finishes"
The handoff to the buyThe funded option list with its profiles reaches the buy without re-entry"Commit this assortment. What does the buyer see, and what did they have to re-key?"
Optimization honestyClear answers on history, new options and overrides"What does your model do with an option that has never sold?"

Two further criteria apply to any planning software and bear repeating here because the assortment layer makes them acute. Who changes the hierarchy, the cluster definition or the attribute set a year after go-live, and what it costs, decides whether the system keeps up with the business; the implementation model guide prices it. And how the sell-through gets in, from which systems and how often, decides whether the in-season job is possible at all; the selection guide's sixth step covers data connectivity. A brand not yet ready for a system can run the same arithmetic in the assortment planning template, which holds the option count and size curve arithmetic in workbook form; the signals that the workbook has been outgrown are the three described at the start of this guide, the envelope moving, the line moving and the grid being re-cut for a second channel, happening in the same season.

What the software has to handle natively, by vertical

The seven jobs are the same in every retail vertical. What differs is the object the option is, what depth physically means, the clock the season runs on, and the constraint that binds the plan, and a system that does not carry a vertical's grain natively pushes the planner back into a spreadsheet beside it, which recreates the problem the system was bought to solve. Assortment planning by vertical covers the discipline in each of these ten; the paragraphs below say what the software has to hold.

Apparel

Apparel plans at style, commits at style-color and sizes at style-color-size, across two or more seasons a year with delivery drops inside each, and, where the brand sells through more than one channel, splits the assortment across DTC, wholesale and stores. The software has to hold the style-color matrix natively, carry a size curve by class and cluster, plan options into delivery drops rather than into a season as a whole, keep channel grids inside one plan, and roll the plan up by collection and fabrication as well as by class. Fabric minimums bind the colorway count upstream of the buy, so the system has to show when a colorway's planned depth across all its channels falls under the mill's minimum; planning against fabric minimums covers that constraint. Color depth, the number of colorways per style, is the apparel-specific breadth decision, and the apparel brands page sets out the apparel case.

Footwear

Footwear multiplies the unit. A model-color is bought as a size run across sizes, half sizes and, where the model offers them, widths, so the option's depth is a run, and the presentation minimum is size-run completeness: the core sizes present in every door, the full run in the doors that can hold it. The software has to plan pairs by size within width, hold the run as the unit of depth so that a cut to depth is a decision about which sizes to drop rather than a flat percentage, keep the run intact through the grid and the buy, and carry the model year as a planning dimension beside the season, because carryover models continue across seasons and the changeover is an assortment decision with its own exit. Assortment planning for footwear brands covers the run arithmetic; see also the footwear brands page.

Accessories and bags

Accessories plan on style-colorways rather than size runs, so the option's multiplier is the colorway and the hardware finish, and much of the range is unsized. The assortment is layered: an evergreen core that carries over season to season and replenishes on a rule, a seasonal layer bought once around a hero color, and a wholesale range that accounts book from. The software has to hold colorway and finish as attributes the plan rolls up by, carry carry-forward as a status with its own planning logic rather than as a tag, plan the hero color's depth against the colorway's minimum at the tannery or component supplier, and read attach rate, the rate at which an accessory sells with a garment or a shoe, as an input to depth. Planning accessories lines covers the hero-color arithmetic; see also the accessories brands page.

Home and furniture

Home and furniture plan SKUs with long lead times, committed a season or more ahead and shipped by ocean in containers whose minimum can set the buy rather than the option's demand, and configured as options, a frame in several finishes, fabrics or sizes, that multiply the SKU count without multiplying the floor sample. The assortment decision is as much about the floor set as about the range: which options are shown as vignettes, which are stocked, which are special order against a sample. The software has to carry option and finish as planning grain, hold the container cube and minimum as a constraint on depth, plan the floor set as the decision about which options are presented in each store grade, hold special-order options as carried but unstocked, and net dealer prebooks written at market against the buy. Merchandise planning for home and furniture brands covers the mechanics; see also the home and furniture brands page.

Outdoor

Outdoor plans on the model year rather than the season alone, with a wholesale channel that prebooks the line at a sales meeting months before delivery, so the assortment is set twice: once as the range dealers can book from, and again as the production buy sized from the prebook with at-once stock held back for fill-in. The line mixes soft goods sized like apparel with hard goods that have sizes of their own or none, so the software has to hold two option grains in one plan, carry the model year with carryover models continuing across it and the changeover planned as an exit, hold the dealer prebook position against the buy, and plan the DTC and wholesale assortments for the same model on different clocks. Merchandise planning for outdoor brands covers the model-year rhythm, and planning a model year changeover covers the exit; see also the outdoor brands page.

Sporting goods

Sporting goods run several seasons at once, one per sport, each with its own start, peak and end, so the assortment is a set of sport-specific ranges that share a store but not a calendar, and the prebook from team and institutional accounts sits beside retail demand for the same items. The software has to carry seasonality by sport so that each range's option count and depth are planned against that sport's window, hold prebooks by sport season as committed demand the buy nets against, plan hardgoods with their own size systems, racquet grips, bat lengths, boot sizes, beside softgoods with apparel sizing, and read the in-season chase against a window that does not move for a sport that starts on a fixed date. Merchandise planning for sporting goods brands covers the multi-season structure; see also the sporting goods brands page.

Beauty

Beauty plans shade ranges, a foundation in dozens of shades, a lip color in a dozen, where the option is the shade and the set has to stay complete, because a missing shade reads to the customer as the product being unavailable. At retail partners the listing is planogram-driven: the retailer's reset calendar and facing count decide when a range can change on shelf and how many shades it can hold, so the assortment grid is set door by door against a planogram the brand does not own. The software has to plan the shade ladder as a set, moving depth within it rather than cutting shades out of it, hold the retailer's gondola reset dates as the calendar the assortment changes on, carry facings per door as the fixture capacity, read shelf-life dating, less the remaining life the retailer requires at delivery, from the batch system as a ceiling on depth for slow shades, and run replenishment natively for the part of the range that is continuous. Merchandise planning for health and beauty brands covers shade-range planning; see also the health and beauty brands page.

Toys and games

Toys concentrate demand in a gifting peak at the end of the year, so the assortment is planned against a few weeks, bought far ahead of them with little chance to chase, and presented at retail partners in planogram facings filled in case multiples. Licensed product runs in windows tied to a film or a franchise, with a start and an end the brand does not control and an exit that has to be planned into the buy. The software has to phase the plan around the peak, hold the licensed window as a planning attribute with its own start, end and exit, plan depth in case packs so that a door's facing is filled in whole cases, carry the property as an attribute the mix rolls up by, and plan the post-peak exit as part of the pre-season plan rather than as a January surprise. Merchandise planning for toy and game brands covers the peak, and planning a licensed product window covers the window; see also the toys and games brands page.

Baby and juvenile

Baby and juvenile demand is driven by the registry and the due date, so a part of the demand for a car seat, a crib or a stroller is visible months ahead in a way no other vertical enjoys, and a part is replenished on a steady rate because the product is consumed as the child grows. The option is often a pattern or a colorway on a certified product whose base cannot change without re-certification. The software has to read registry demand as a planning input to depth and to the grid, run replenishment on core items natively, carry stage ranges rather than body sizes as the size dimension, plan patterns on a certified base as options with a shared constraint, and absorb a safety recall, which can remove an option from sale in a day and leaves a hole in the grid, the open-to-buy and the registry promise at once. Merchandise planning for baby and juvenile brands covers registry demand, and planning with registry demand covers the mechanism; see also the baby and juvenile brands page.

Jewelry and watches

Jewelry and watches plan the piece in a metal, often placed door by door rather than cluster by cluster, on a precious-metal cost base that moves between the plan and the sale, so the margin the assortment is planned on has to be re-read against the metal price, and the turn is slow, with pieces carried across seasons and model years rather than cleared at the end of one. The software has to carry metal and stone content as cost drivers on the option, plan a model year with carryover as the default and the changeover as a planned exit, hold low depth per piece across many pieces without reading slow turn as a markdown signal, support door-level placement where a piece's presence in a door is a security and a presentation decision as well as a demand one, and plan the gifting peaks that concentrate demand. Merchandise planning for jewelry and watch brands covers the cost base, and planning margin on a moving cost base covers the re-read; see also the jewelry and watch brands page.

The pattern across the ten is the one the merchandise planning software guide names: the vertical sets the grain, and the software either carries that grain natively or pushes it back into a spreadsheet. The evaluation test that follows from it is to plan the brand's own object in the demonstration, pairs by size within width, a shade ladder as a set, a frame in four finishes, a licensed window with its exit, before any other feature is looked at.

Common failure modes

Implementations of assortment planning software fail in a small number of repeating ways, and each one has a mechanism. The list below states the mechanism rather than the frequency, because the frequency is not measurable and the mechanism is what a buyer can check for.

  • Planning in SKUs before the line is locked. The plan is built on real style numbers from the first pass, so it cannot start until design has drawn the line, which is too late for the financial frame to shape the line, and every style that changes in development breaks the plan. The fix is placeholder planning: slots with a tier, a depth and an attribute set, filled as the line becomes real.
  • Grids with no presentation minimums. The option count is set from the envelope and the depth per option is what is left, with no floor per door, so the plan reads as fully funded while the depth in each door has fallen below the quantity that makes the option visible. The customer sees a gap where the option should be, and the hindsight reads it as a weak option rather than as an under-presented one.
  • Breadth the buy depth cannot fund. The merchandiser adds options to cover every story and every price point, each one shaving a little depth from the others, until every option is below its minimum viable depth and the range is broad and under-supported everywhere. The mechanism is that shaving looks cheaper than cutting; the fix is to test every option against its floor and to cut the ones that fall under it, which frees depth for the rest.
  • Optimising history on a line that is mostly new. The optimization engine is run on a season where, in an illustrative split, six options in ten have no history, and its confident numbers for the new options are read as measurements. The fix is to read the model's output as a baseline for continuing options and as an analogue for new ones, and to keep the product judgement in the loop where the analogue is weak.
  • Attributes as text. The product record carries silhouette, fabrication and color family as free-text columns, so the roll-ups that make the attribute balance job possible cannot be computed, and the mix is judged by eye. The fix is attribute dictionaries maintained once, which is a data-readiness task before it is a software one.
  • The grid built after the buy. The buy is committed at total depth and the split by cluster and channel is decided afterwards by the allocator, so the depth per door is whatever falls out, and the presentation minimum is met in some doors and not others. The fix is to build the grid in the assortment plan, before the buy, so that depth per option is the sum of the cells rather than a total to be divided.
  • Two plans, one season. The assortment is planned in the system and re-planned in a workbook beside it, because the system cannot carry the vertical's grain or the brand's cluster definition, and the workbook becomes the plan of record while the system holds a copy. The mechanism is a grain mismatch; the fix is to test the grain in the evaluation.
  • The in-season read in a different file. Sell-through comes back into a reporting layer that does not hold the plan, so the reassortment and chase decisions are made against a dashboard and the plan is never updated, and the next season's hindsight compares actuals with a plan that was abandoned in week three. The fix is the seventh job: the system that holds the plan reads the actuals against it.

A worked example: option count and depth from an OTB envelope

The arithmetic below is the first job in the list above, done by hand for one class, one price tier, one cluster and one delivery. 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, and the fact that a planning system has to redo it every time any input moves.

The inputs:

  • The open-to-buy allows planned receipts of $288,000 at retail for women's knit tops in the better price tier, for cluster A, in the first spring delivery.
  • The better tier's planned ticket price is $60, and receipts are planned at ticket, so the envelope is 4,800 units ($288,000 divided by $60).
  • Cluster A has 40 doors, all of which carry the full size range.
  • The presentation minimum for an option in this class is one complete size curve per door: XS 1, S 2, M 3, L 3, XL 2, XXL 1, which is 12 units. Below that a door is missing sizes on the first day.
  • The plan holds a replenishment reserve of 6 units per door per option, half a curve weighted to S, M and L, to refill the core sizes once during the delivery's weeks on the floor.

The derived figures:

LineCalculationResult
Envelope in units$288,000 at $604,800 units
Floor set per option40 doors × 12 units480 units
Reserve per option40 doors × 6 units240 units
Planned depth per option480 + 240720 units
Options fully funded4,800 ÷ 720, rounded down6 options
Units used by six options6 × 7204,320 units
Remainder4,800 − 4,320480 units

Six options at 720 units each use 4,320 of the 4,800 units and leave 480. The remainder is exactly one floor set, and that is where the breadth-versus-depth decision lives.

Scenario A, breadth: carry a seventh option at floor depth only. It is present in all 40 doors with a complete curve on day one, and it has no reserve, so when its core sizes sell it is broken until the next delivery. The plan is six options at 720 and one at 480: 4,320 + 480 = 4,800 units, which at $60 is $288,000, the whole envelope.

Scenario B, depth: carry six options and give the 480 remaining units to the two options planned as the delivery's heroes, 240 each, doubling their reserve. The plan is two options at 960 and four at 720: 1,920 + 2,880 = 4,800 units, again $288,000.

Both scenarios spend the envelope to the unit, so the open-to-buy cannot choose between them; the choice is a merchandising judgement about whether the delivery needs a seventh story more than the heroes need a second refill. What the software contributes is that both scenarios exist at once, each with its option count, its depth per option, its spend and its attribute roll-up, and that the seventh option in scenario A is flagged as floor-only rather than looking like the other six.

The what-if that shows why the system has to hold the relationships: suppose the cluster is redefined and four small doors move to cluster B, leaving 36. The floor set per option becomes 36 × 12 = 432 units, the reserve 36 × 6 = 216, and the planned depth 648. The same 4,800 units now fund seven options at 648 (7 × 648 = 4,536) with 264 left over, and 264 is below the 432 floor, so it cannot fund an eighth option and goes to reserve on the heroes instead. One change to the cluster definition moved the option count from six to seven and changed every depth in the plan. In a workbook that is a rebuild of every tab the cluster touches; in a planning system it is the same plan with one input changed, which is the whole case for the category.

Where RetailNorthstar fits

RetailNorthstar's assortment planning page describes the capability in these terms: build seasonal assortments aligned to financial guardrails, with assortment decisions connected directly to the OTB budget and buy plan and no manual reconciliation. It lists attribute-level planning, by style, color, size and fabric with automatic rollup to department totals; OTB-constrained assortment, in which assortment decisions reconcile against the OTB budget in real time; channel segmentation, with separate assortments for DTC, wholesale and marketplace and channel-specific targets; visual assortment on Canvas, where products are dragged onto category, delivery, channel or tier boards and OTB, option counts and mix update as they are placed; performance-linked recommendations, in which AI surfaces product candidates based on historical sell-through and margin contribution; and buy plan auto-generation, converting final assortment decisions into structured buy plans with one action. On Canvas, the page says, merchandisers draft the assortment by placing products onto boards, review it as a team and commit it in the assortment meeting, and the committed board is the assortment plan. Visual assortment planning describes that workflow in more detail.

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 category, the cluster definition and the open-to-buy figure, and build the plan in the demonstration. Which 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

See how RetailNorthstar connects assortment decisions directly to the OTB budget and buy plan — OTB, option counts and mix update as you place products on Canvas.

Start Free Trial →

No credit card. No commitments.

Common questions

What is assortment planning software?

Assortment planning software is the system a brand or retailer uses to decide which options to carry in a season, for which channels and store clusters, and how deep to buy each one by size and pack, inside the open-to-buy that funds them, holding that plan as one connected object from the first placeholder to the committed buy. It is the system of record for the assortment plan: the option count by class and price tier, the depth behind each option, the grid that says which cluster and channel carries it, and the size and pack profile the buy will be cut to.

What does assortment planning software do that an assortment spreadsheet does not?

It holds the relationships between the inputs rather than one state of them. The option count depends on the depth, the depth on the presentation minimum and the door count, the door count on the cluster definition, and all of it on an open-to-buy figure that moves. A workbook holds one version of that system and every change is a rebuild; a planning system re-derives the others when any input changes, holds scenarios side by side, keeps the financial frame attached as placeholders become styles, and reads the in-season actuals against the same plan it was built in.

How is assortment planning software different from merchandise planning software?

Assortment planning software is one layer of merchandise planning software. Merchandise planning software covers the whole stack, the financial plan, the open-to-buy, the assortment, the buy, size and pack, allocation and the in-season reforecast, on one data model. Assortment planning software does the assortment layer: option count and depth against the open-to-buy envelope, grids by cluster and channel, attribute and good-better-best balance, size and pack profiles, placeholder-to-style evolution, what-if on breadth versus depth and in-season reassortment. A brand can buy the layer as a point solution and integrate it with whatever holds the rest, or buy a connected platform that holds the assortment as one module beside the others.

What is assortment optimization, and when does it work?

Assortment optimization is a set of models that choose or adjust options from a candidate set on the basis of history: demand transfer estimates what happens to an option's demand when it is removed, like-item modelling borrows the performance of similar past options as the baseline for a new one, and cannibalisation estimates what adding an option takes from the ones already planned. The mechanism needs options with history and a similarity measure that describes what the customer responds to, so it works on continuing items, replenished categories, carryover programs and grids being localised on the basis of how the same items sold in each cluster. It degrades on fashion newness without history, where its confident numbers for new options are analogues that should be read as a baseline for a product judgement, not as a measurement.

What data does assortment planning software need to hold?

A merchandise hierarchy from division to SKU; product attributes held as planning dimensions rather than text; store clusters and grades with their door counts and capacities; channels; price tiers with target price points and margin; a ladder from placeholder to style to style-color to SKU; presentation minimums as a floor on depth per option per door; fixture capacity as a ceiling on options per cluster; the calendar with its deliveries and, where the vertical uses it, the model year; the open-to-buy envelope by class, tier, channel and period; size curves and pack configurations; and actuals by SKU by door by week. The three to check for first are the constraints: presentation minimums, fixture capacity and the placeholder ladder.

How does assortment planning software connect to OTB, line planning and allocation?

The open-to-buy is the envelope above it: the assortment spends the receipt budget option by option and every change to depth moves the position, so the software reads the envelope as a live constraint and pushes committed depth back as spend. The line plan is the menu beside it: line planning decides what could exist and assortment planning decides what is bought, so placeholders in the assortment become styles from the line, and where the line is planned on visual boards the seam is cleanest when the board and the plan are the same object. Allocation is below it: the assortment grid by cluster and channel is the target the allocator distributes within, so the grid has to be built in the assortment plan before the buy rather than divided afterwards.

What should a brand ask a vendor of assortment planning software?

Bring your own category, cluster definition and open-to-buy figure and ask the vendor to build the plan in the demonstration. Then test the seams and the constraints: revise the envelope while the grid is open and see what changes; build a class on placeholders and replace some with real styles; add an option the envelope cannot fund at the presentation minimum and see what the system does; put more options into a small-door cluster than its fixtures hold; change an option from open stock to prepacks and watch the depth; hold two versions of the plan side by side; ask what the optimization model does with an option that has never sold; and commit the assortment and ask what the buyer had to re-key.

Does the same assortment planning software work for footwear, beauty, furniture and other verticals?

The seven jobs are the same in every vertical; the grain is not. Footwear plans pairs by size within width and needs size-run completeness as the presentation minimum and the model year as a dimension. Beauty plans the shade ladder as a set against a retail partner's planogram and reset calendar. Home and furniture plan a frame in several finishes against container minimums and a floor set of vignettes. Toys plan licensed windows and case-pack facings around a gifting peak. Jewelry plans the piece in a metal on a moving cost base, placed door by door. A system that does not carry the brand's own grain natively pushes the planner back into a spreadsheet beside it, so the evaluation should plan the brand's own object before any other feature is looked at.

RetailNorthstar Editorial Team
RetailNorthstar ·

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.

Take the assessment →

Apply these insights with RetailNorthstar.

See how modern apparel brands use RetailNorthstar to put this planning framework into practice.

No credit card. No commitments.

Connected merchandise planning — live in weeks, not quarters.