What Is Merchandise Planning? The Five Plan Layers and How They Hand Off

What is merchandise planning?

Merchandise planning is the discipline of deciding what a brand will sell, how much inventory to own in order to sell it, and where that inventory should sit — expressed as a stack of five connected plans, each denominated in a different unit and each constraining the one below it. The layers run from the merchandise financial plan in dollars, through the line plan in options, the assortment plan in style-colors, the buy plan and open-to-buy in units and committed dollars, to the allocation plan in units by location.

The common definition stops at the first sentence and describes merchandise planning as "the process of planning what to sell." That is true and it is useless, because it does not tell a planner what to produce on Monday. The useful version of the definition is structural: merchandise planning is a translation chain. Each layer takes the previous layer's currency and converts it into a more specific one, and every real decision in the cycle happens at one of those conversion points.

That framing matters for a practical reason. When a season goes wrong, the symptom surfaces in the last layer — wrong sizes in the wrong doors, a markdown that was not budgeted, a chase that could not be funded. The cause is usually upstream, because upstream is where the conversion was made: a layer where the translation was done loosely or skipped entirely. The only way to find it is to know what each layer was supposed to decide.

This guide maps the five layers: what each one decides, what unit it is measured in, what artifact it produces, who owns it, how often it runs, and what specifically breaks downstream when it is skipped or done badly. It then covers three things the layer map alone does not settle — where merchandise planning stops and other disciplines begin, how the layers compress in a brand where one person does all five, and the reconciliation problem between top-down dollars and bottom-up units.

The five layers at a glance

The table below is the structure the rest of this guide expands. Read the unit of measure column first: it is the clearest single indicator of which layer a conversation is actually in. A discussion in dollars is layer one. A discussion about how many options a class should carry is layer two. A discussion about whether a color should ship to the outlet cluster is layer three.

Every cell in that column is denominated against the merchandise hierarchy — department, class, style-color, SKU — so a stack written on an unstable hierarchy cannot roll up. What occupies each hierarchy level changes completely by category, and the level names for apparel, footwear, accessories and bags, home and furniture, outdoor, beauty, sporting goods, toys, baby and juvenile, and jewelry and watches are set out in merchandise hierarchy by vertical.

Two things are worth noticing before the detail. First, the owner column changes hands three times across the stack, which is why the handoffs are where plans die — each one crosses a functional boundary. Second, the cadence column is not uniform: layers one, four and five keep running all season, while layers two and three effectively stop at commit. A brand that treats all five as pre-season exercises loses the in-season half of the discipline entirely.

Layer 1: The merchandise financial plan

The merchandise financial plan owns one decision: how much inventory the business can afford to own, month by month, in order to support a planned level of sales at a planned margin. It works exclusively in dollars, usually at department or class level, and it is the only layer where the numbers tie directly to what finance sees.

Its artifact is the seasonal MFP — a grid of months against plan lines: sales, markdowns, receipts, beginning-of-period and end-of-period inventory, gross margin. Those lines are not independent. They are bound by the inventory identity, which is what makes the MFP a plan rather than a wish list: ending inventory equals beginning inventory plus receipts minus sales minus markdowns and shrink. Change any one line and at least one other must move. The full mechanics of building it — the plan lines, the identity, the seasonal build sequence — are covered in the merchandise financial planning guide.

The MFP is built pre-season and then reforecast on a monthly rhythm, because it is the layer that absorbs actual performance. When sell-through runs ahead or behind, the receipt and markdown lines are where the correction lands first.

What breaks downstream when this layer is weak. Without a credible MFP, the assortment has no ceiling. The buy sums to whatever the line wants to be, which is always more than the business can carry, and the excess is discovered at receipt rather than at plan. The characteristic symptom is a markdown budget that does not exist as a plan line and therefore appears as a surprise in the second half of the season, spent through markdown optimization decisions nobody funded — the money was always going to be spent, it simply was never planned. A second, quieter symptom is a receipt flow that is correct in total for the season and wrong in every individual month, because nobody phased the dollars against the trading calendar before handing them down.

If you want to see the shape of this layer before building one, the merchandise plan template lays out the same rows — sales, markdowns, inventory, receipts — against a season.

Layer 2: The line plan

The line plan owns a decision that is easy to leave implicit and expensive to get wrong: how many options should exist, in which classes, at which price points, arriving in which delivery. It is denominated in counts — option count, styles, sometimes carryover versus new — and in the price architecture those options sit within.

Its artifact is the line plan itself: a grid of classes against option counts, with target retail price bands, target initial markup, and a delivery or flow assignment. It is owned by merchandising and built in conversation with design and product development, because it is the layer where a commercial constraint meets a creative process. Some brands file line planning under design or product development instead, and where the ownership sits is a genuine convention difference rather than an error — but the sequencing is not negotiable either way: the option count has to be set before the line is designed, or it becomes a count of what was designed. Line planning normally locks at line freeze, after which adding an option is expensive in sampling and lead time rather than merely in dollars.

The critical property of a line plan is that the line is always larger than the assortment will be. It defines what could exist; the next layer decides what will actually be bought. That distinction is the source of most confusion between these two layers, and it is worth reading the dedicated treatment in line planning vs. assortment planning if the two are currently run as one exercise.

What breaks downstream when this layer is weak. Option count expands to fill available enthusiasm. Because the receipt dollars are fixed by layer one, more options means less depth per option — so every style is under-supported, size runs break early, and the brand ends the season having produced a wide line that never had enough of anything. The second failure is duplication: two options in the same class at the same price point in adjacent colorways split the same demand between them, and both look like underperformers in the post-season review. Neither failure is visible in the buy, which is why it has to be caught here.

Layer 3: The assortment plan

The assortment plan owns the decision the term is easiest to picture: which specific style-colors go to which channels and door clusters, and at what depth. It moves the unit of measure from options to style-colors, and it is the first layer where the size curve enters the arithmetic.

Its artifact is the assortment grid — the style-color matrix, with style-colors down the rows, channels and door clusters across the columns, and planned depth in the cells. In a wholesale-heavy business the same grid is expressed as a line sheet with account-level segmentation. In a DTC-only business the cluster dimension may collapse to one column, which makes the layer look simpler than it is; the depth and size-curve decisions are unchanged. The full process, including how breadth and depth trade against each other, is in the assortment planning guide.

This is also the layer where a brand's differentiation actually gets expressed. A single assortment sent everywhere is not an assortment plan, it is a catalog. The plan exists to make deliberate differences between clusters — a warmer color story in one region, a deeper size run where the demographic supports it, a shallower buy on a fashion option in the doors that historically clear it late.

What breaks downstream when this layer is weak. The buy gets placed at style level with no channel or cluster differentiation, so every door receives the same mix regardless of what it sells. The visible failure is simultaneous stockout and overstock on the same style-color in the same week across different doors — a total that looked correct in aggregate and was wrong everywhere in detail. The second failure is size brokenness: without a size curve applied at the style-color level, the buy is placed on a blended curve, and the doors with an atypical size profile go out of stock in their core sizes while holding the tails.

Layer 4: The buy plan and open-to-buy

This layer owns the conversion from intention to commitment: how many units to actually purchase, at what cost, arriving in which month, and how much budget remains uncommitted. It is denominated in units and in committed dollars at cost, and it is the point at which a plan becomes a liability on a purchase order.

It carries two related artifacts. The buy sheet or PO plan lists what is being committed, to which vendor, at what cost, for which delivery. The open-to-buy is the running control on top of it — the calculation of how much of the receipt budget is still available after existing commitments. The canonical form is straightforward:

Open-to-buy is the one artifact in the stack that is genuinely alive all season. It is reviewed weekly in-season because every cancellation, chase order and markdown changes it. A brand with no open-to-buy discipline does not have a smaller version of this layer — it has no version of it, because commitment without a remaining-budget calculation is just a sequence of purchase orders. The mechanics, and the places OTB discipline breaks down in practice — spreadsheet disconnection, stale on-order data, over-aggregated buckets — are covered in the open-to-buy planning guide.

What breaks downstream when this layer is weak. Commitments quietly exceed the receipt budget, and because purchase orders land over several months the overage is invisible until the month it arrives. A consequence with no recovery path is the loss of chase capacity: a brand that has committed its entire budget pre-season has nothing left to buy into the styles that are actually selling, so its best performers go out of stock while its worst ones arrive on schedule. The other failure is phasing — receipts that are correct for the season but land after the weeks they were meant to sell in, which converts a good buy into a markdown.

For the same stack framed for a leadership audience rather than a practitioner one — what to ask for, and what good looks like at each layer — see the executive guide to AI-assisted apparel merchandising.

Layer 5: The allocation plan

The allocation plan owns the last conversion: which units go to which location, in which sequence, and in what size distribution. Its unit of measure is units by location and size, and its artifact is the distro or allocation run attached to each delivery.

Allocation is where a well-made plan is either realized or wasted, and it is the layer most often treated as an administrative step rather than a planning decision. It is not administrative. Deciding whether to push a full initial allocation or hold back a warehouse reserve for in-season reallocation is a genuine strategic choice with a real trade-off: pushing everything maximizes floor presence and minimizes handling, while holding reserve preserves the ability to respond to actual sell-through at door level. The practical patterns, including how initial allocation differs from ongoing replenishment, are in the allocation and replenishment guide.

This layer is also where the planning stack meets the operations chain that has to physically deliver it — production, work in progress, inbound logistics and the receiving calendar. Those seams, and what breaks at each of them, are covered in mastering apparel operations.

What breaks downstream when this layer is weak. The brand buys the right total units and puts them in the wrong places. Aggregate sell-through looks mediocre; door-level sell-through shows a wide spread with some locations stocked out in week three and others still holding the same style at season end. Because the aggregate number is the one that reaches the post-season review, the diagnosis usually lands on the buy — and the following season's plan corrects a layer that was never the problem. The tell for an allocation failure rather than a buy failure is variance between doors, not the level of the average.

The handoff numbers that actually matter

The artifacts described above are documents, and documents are not what makes the stack work. What makes it work is that each layer emits one or two specific numbers that the next layer is not allowed to contradict. Those numbers are the real interface, and a planning process can be judged almost entirely on whether they exist and whether anyone checks them.

Read that table as a loop rather than a line. The fifth row closes it: actual performance at door level is the input to the next reforecast of layer one, which is what turns the stack from a pre-season exercise into a working cycle. A planning process that runs the first four rows and never closes the loop will build next season's plan from last season's plan rather than from last season's results.

Where merchandise planning starts and stops

The term is used loosely enough that it is worth naming the boundaries directly. Each of these is a real seam, and none of them is a clean wall.

It is not demand planning. Demand planning estimates what customers will buy; merchandise planning commits to what the business will own and what it must return. The seam sits at the buy quantity, and the three-way separation between demand, merchandise and supply — including where supply planning sends work back upstream and why the same words mean different things in grocery than in apparel — is worked through in demand planning vs merchandise planning vs supply planning.

It is not supply planning. Merchandise planning names the delivery date it needs; supply planning decides whether that date is achievable. The purchase order is the boundary object, and the seam leaks in one direction reliably: minimum order quantities and case-pack constraints from the supply side override depth decisions made in layer three, which is why those constraints belong in the assortment conversation rather than being discovered at buy.

It is not replenishment. The allocation plan is an initial push against a planned assortment. Replenishment is an in-season pull that responds to sales on continuity or core product. The boundary is genuinely fuzzy for never-out programs, where layer four is not planning a one-time buy at all but a flow with a reorder rhythm — for those items the buy plan and the replenishment logic are the same object viewed at two horizons.

It is not product development or design. The line plan sits exactly on this seam. It says a class needs a certain number of options in a certain price band arriving in a certain delivery; it does not say what those options look like. When the line plan is written after design rather than before, the layer has been inverted and the option count becomes an output rather than a constraint.

It is not pricing or markdown optimization, though the markdown budget lives in layer one and the price architecture lives in layer two. The in-season decision of which style-color to mark down by how much is a separate discipline that spends the budget merchandise planning set.

The reconciliation problem

Top-down dollars and bottom-up units do not agree on the first pass. This is not a sign that something has gone wrong. It is the expected result of two plans built from different information: the top-down plan starts from a growth target and last year's shape and is expressed in department dollars, while the bottom-up plan is assembled style-color by style-color. The bottom-up number is larger by construction: it is built from what merchandising believes it can sell, not from what the envelope allows, and nothing in the way it is assembled forces it to stop at the envelope.

The failure is not the gap. The failure is closing the gap late, or closing it by silently letting one side win. A gap resolved after receipts are committed has exactly one remaining lever, and that lever is markdown.

Here is an illustrative reconciliation, with round numbers chosen for clarity rather than drawn from any brand's actuals.

Suppose a department's merchandise financial plan allows 900,000 dollars of receipts at cost for the season. The bottom-up assortment plan carries 60 style-colors at an average planned depth of 500 units, at an average unit cost of 32 dollars. That is 60 times 500, or 30,000 units, at 32 dollars each — 960,000 dollars. The plan is over by 60,000 dollars, which on these numbers is close to four options' worth of buy: 1,875 units that have to come out of the plan somewhere.

Only the option cut clears 60,000 dollars on its own, and it overshoots by 4,000. The depth cut recovers 38,400 and the retimed delivery 32,000 — each closes roughly half to two-thirds of the gap, so either has to be combined with something else. Cutting depth by 20 units and retiming one delivery together recover 70,400, which clears the gap with room; a smaller version of each would land closer to it. That combination arithmetic is the normal shape of a reconciliation — one lever rarely lands on the number, and which levers get combined is the actual decision.

They are not equivalent decisions. Cutting options protects depth and therefore size integrity, at the cost of breadth. Cutting depth protects breadth and puts every option at higher risk of breaking size early. Retiming protects both but moves the problem into the next period, where it is only a real solution if the next period genuinely has room. Raising the envelope is legitimate when the original margin or turn assumption was wrong, and is self-deception when it is used to avoid choosing.

The reconciliation itself is a meeting, not a spreadsheet function, because the choice between those levers is a merchandising judgment. What the spreadsheet must supply is the gap, expressed in both currencies, before the buy is committed. A step-by-step reconciliation sequence — setting top-down targets first, building bottom-up by category, then locking and tracking the resolved plan — is laid out in aligning top-down and bottom-up planning.

Cadence, and why the layers run on different clocks

Every layer in the stack is denominated in the retail calendar, and most of them are phased by week rather than by month. That matters more than it sounds: monthly figures inside a quarter are not comparable to each other under the 4-5-4 family of conventions, because one month in each quarter carries five weeks rather than four. Anyone phasing a receipt plan by dividing a quarter into three equal months has misphased it before the first buy. The conventions and the traps are covered in the retail calendar guide.

The layers also run at different frequencies, and that mismatch is the source of most friction between them. Layers two and three are seasonal and effectively stop at commit. Layer one reforecasts monthly. Layers four and five run weekly all season. A change made in a weekly layer routinely invalidates an assumption in a seasonal one, and because the seasonal artifact is no longer being opened, nobody notices. This is the mechanical reason plans and reality diverge without anyone making a mistake: the fast layers keep moving, the slow ones do not, and the connection between them is a spreadsheet nobody has reopened since line freeze.

How the layers compress in a small brand

In an emerging brand, one person does all five layers, often in one workbook, often in a single week. The layers do not disappear when the roles do. What compresses is the artifacts; what does not compress is the sequence of decisions.

The specific failure in a compressed process is that the option count stops being a decision and becomes a discovery. Instead of setting an option count from the receipt budget and filling it, the planner builds an assortment from what design produced, totals it, and finds out what it costs. That is the same stack run backwards, and it produces the same symptoms as skipping layer two entirely: too many options, too little depth, and a gap discovered at commit.

A workable minimum for a small team is five objects, not five documents:

• →A one-page seasonal financial plan by month, carrying sales, markdowns, receipts and end-of-period inventory, with the inventory identity actually enforced by formula.
• →An option count by class and price band, set from the receipt dollars before the assortment is built.
• →A style-color grid with a size curve, differentiated by channel even if there is only one door cluster.
• →An open-to-buy that shows committed versus open dollars by month, updated whenever a purchase order is placed.
• →A distribution rule, even a simple one, written down rather than decided per delivery.

Keeping these as linked tabs in one workbook rather than five separate files is usually the right call at small scale, on one condition: the numbers that cross between layers must be formulas, not typed values. A retyped receipt budget is how a compressed plan loses its connection to the financial plan it was supposed to obey: the number stops recalculating, so the layer above can change without the layer below ever hearing about it. When the option count no longer recalculates from the receipt dollars, the stack has quietly become five unrelated worksheets that happen to share a filename.

The same principle scales. Whether the five layers live in one workbook or five systems, the discipline is identical: know which layer you are in, know the number it owes the next layer, and check that the number still holds after anything changes.