Merchandise Planning Explained: The Complete Guide for Fashion Retail

Merchandise planning is the process of deciding what to buy, how much to buy, where to put it, when to move it, and when to mark it down. Done well, it is the single biggest driver of profitability in modern retail. Done poorly, it is the reason retailers end every season with racks of unsold inventory and margins thinner than planned.

This guide covers the full merchandise lifecycle, from pre-season planning through end-of-season management, with specific attention to the challenges and opportunities facing multi-branch modern retailers in South Africa.

What is merchandise planning?

At its core, merchandise planning answers six questions:

1. What should we buy? (Assortment planning)
2. How much of each should we buy? (Open-to-buy and quantity planning)
3. Where should it go initially? (Allocation)
4. Where should it move during the season? (Rebalancing and replenishment)
5. What price should it be? (Pricing and markdown)
6. What happens at end of season? (Returns, clearance, and learning)

These questions form a lifecycle. Each decision affects every subsequent one. Get the assortment wrong, and no amount of rebalancing will save you. Get the allocation wrong, and you are rebalancing all season. Fail to rebalance, and you are marking down stock that could have sold at full price somewhere else.

The most important thing to understand about merchandise planning is that it is not six separate activities. It is one interconnected system. Treating each stage in isolation -- which is what most mid-market retailers do when they manage each function in separate spreadsheets -- guarantees suboptimal outcomes.

The merchandise lifecycle

Stage 1: Assortment planning

Assortment planning happens before the buying season. It determines the range: which styles, in which sizes and colours, for which store clusters.

Range architecture

Range architecture defines the breadth and depth of the assortment. Breadth is how many different styles you carry. Depth is how many units of each style.

The tension is fundamental: broader ranges offer more choice but spread inventory thinner. Deeper ranges ensure availability but risk overstock on styles that do not perform.

For multi-branch retailers, range architecture also includes store-level variation. Not every store should carry every style. A flagship CBD store might carry the full range at shallow depth. A suburban outlet might carry a curated selection at greater depth. Assortment planning intelligence uses store clustering and demand data to build tailored range plans for each store group.

Size and colour matrix

Within each style, you must decide the size and colour breakdown. This is where most retailers get into trouble, because the decisions are typically made using national average data.

As pilot data revealed, size demand varies significantly across stores. A store whose customer base peaks at size 5 needs a fundamentally different size matrix than one that peaks at size 8. Applying a national curve to both guarantees that both will be wrong.

Colour decisions face similar issues, compounded by the fact that colour preferences change faster than size distributions and are harder to predict from historical data.

Open-to-buy

Open-to-buy (OTB) is the budget framework for purchasing. It specifies how much (in rands or units) can be spent on new inventory, broken down by category, season, and period.

OTB planning requires balancing:

• Sales plan. How much do you expect to sell?
• Margin targets. What gross margin do you need to achieve?
• Inventory targets. What is the optimal closing stock position?
• Markdown budget. How much markdown activity is planned?

Effective OTB planning is tightly linked to assortment decisions. There is no point planning a range that exceeds the OTB budget, and no point setting an OTB that does not support the range architecture.

Stage 2: Initial allocation

Allocation is the moment of truth. It is when planned inventory becomes physical stock on shelves in specific stores. Once stock ships from the warehouse, the cost of fixing a mistake increases dramatically.

The allocation challenge

Consider a footwear retailer with 80 stores receiving a new season shipment of 500 units of a particular shoe style, across 10 sizes. That is 80 x 10 = 800 allocation decisions for a single style. The retailer carries 200 styles. That is 160,000 allocation decisions, made in a compressed timeframe, for a single season.

Most mid-market retailers make these decisions using allocation formulas in spreadsheets: each store gets a percentage of the total based on last season's sales. This approach is fast and simple. It is also systematically wrong, because it does not account for:

• Per-store size curves. As discussed, size demand varies by store. National-curve allocation means every store gets somewhat wrong quantities.
• Stockout bias. Stores that stocked out of popular sizes last season show artificially low demand for those sizes. Allocating based on this biased signal perpetuates the problem. See The Stockout Bias Problem for a detailed analysis.
• Style-store affinity. Some styles sell better in some stores. A sporty trainer might fly in a store near a gym and sit in a store in a business district. Last season's total sales do not capture this nuance if the allocation itself was wrong.
• New store cold-start. A new store has no sales history. How do you allocate to it? Most retailers use a similar store as a proxy, but "similar" is often defined by geography or revenue rather than by demand profile.

Replenify's allocation module addresses each of these challenges through per-store size curve intelligence, stockout-corrected demand signals, and style-store affinity modelling.

Allocation timing and imports

For South African modern retailers importing from Asia, allocation planning starts 3-6 months before the stock arrives. The buying decision was made even earlier. By the time stock hits the warehouse and is ready for allocation, market conditions may have shifted from when the buy was placed.

This is one reason why allocation cannot be a "set and forget" activity. It needs to be informed by the most current demand signals available, even though the product range was locked months ago.

Stage 3: In-season rebalancing

No allocation is perfect. Even the best allocation degrades within weeks as actual demand diverges from forecasts. Rebalancing is the in-season correction mechanism.

Why rebalancing exists

If allocation were perfect and demand were perfectly predictable, there would be no need for rebalancing. Neither condition holds in reality, which means every multi-branch retailer needs a way to move stock from where it is not selling to where it is.

The challenge is doing this profitably. A stock transfer costs money (freight, labour, system updates). It only makes sense if the expected margin gain exceeds the transfer cost. This cost-benefit calculation must happen for every potential move across the entire network -- a combinatorial problem that manual planning cannot solve at scale.

The 3-pass approach

Replenify's rebalancing engine uses a 3-pass architecture:

1. Pass 1: Warehouse to retail. Push central stock to stores with demand. Lowest cost, highest certainty.
2. Pass 2: Inter-store balancing. Move stock between stores with distance-aware routing. Highest volume, most complex.
3. Pass 3: Sweep and consolidation. Consolidate fragmented stock into sellable assortments. Recovers value from broken ranges.

In the pilot, this architecture processed 28,348 inventory snapshots and generated 1,572 intelligent moves in 265 milliseconds, producing R745,898 in measured GP uplift.

Rebalancing frequency

How often should you rebalance? The answer depends on how fast demand shifts and how quickly you can execute transfers. For most multi-branch modern retailers:

• Weekly is the minimum. Monthly is too slow to capture intra-season demand shifts.
• Daily is ideal. It catches stockout risks before they become lost sales.
• Intra-day is possible with Replenify (the engine runs in under 300ms) but typically limited by logistics execution speed rather than computation.

Stage 4: Replenishment

Replenishment covers reordering of continuing lines -- products that can be resupplied within the season. In modern retail, this is typically a smaller portion of the range than in grocery or general merchandise, but it applies to basics, core styles, and replenishable essentials.

The South African replenishment challenge

For imported goods, replenishment lead times of 60-120 days mean that a reorder decision made in March for the winter season may not arrive until June or July -- well into (or past) the selling period. This makes reorder point calculation critical: order too late and the stock arrives when the season is ending; order too much and you carry excess into off-season.

Replenishment intelligence factors in actual supplier lead times, transit variability, current stock positions, and demand forecasts to recommend optimal reorder timing and quantities.

Safety stock in fashion

Safety stock -- the buffer held against demand and supply uncertainty -- is a delicate balance in modern retail. Too little and you stock out of best sellers. Too much and you end the season with excess that must be marked down.

The right amount of safety stock varies by:

• Product lifecycle stage. New-season products need more buffer (demand is uncertain). End-of-season products need less (the risk of overstock outweighs the risk of stockout).
• Store velocity. High-velocity stores need proportionally more safety stock because stockout costs are higher.
• Supplier reliability. A supplier that consistently delivers on time requires less buffer than one with variable lead times.
• Size/colour. Core sizes (6, 7, 8 in women's footwear) may warrant more safety stock than extreme sizes.

Stage 5: Pricing and markdown management

Pricing is where margin is either protected or destroyed. In modern retail, the primary pricing decision is markdown: when to reduce, by how much, and at which stores.

The markdown dilemma

Every markdown represents a trade-off between margin and speed. A smaller markdown preserves margin per unit but clears stock slowly. A deeper markdown clears faster but at greater margin cost.

The optimal strategy is not one-size-fits-all. It depends on:

• Remaining season. More time left means you can afford smaller, staged markdowns. Less time means you need to be aggressive.
• Stock depth. Deep stock requires steeper markdowns to clear. Shallow stock may sell through with a modest reduction.
• Store-level demand. A style that is dead in one store might still be selling at full price in another. Marking down network-wide when only some stores need it destroys unnecessary margin.
• Competitive context. What are competitors doing? In a price-sensitive market like South Africa, competitive markdown timing matters.

Price optimization addresses this by modelling markdown timing and depth at the store level, respecting margin floors and coordinating across the network.

Why store-level pricing matters

Most mid-market retailers run network-wide pricing: a markdown at one store means a markdown everywhere. This simplifies operations but wastes margin. If a shoe is selling well at Store A (no need to mark down) but sitting at Store B (needs a markdown), a network-wide markdown sacrifices full-price sales at Store A to clear stock at Store B.

The alternative -- rebalancing stock from Store B to Store A before marking down -- is often more profitable. This is why rebalancing and pricing are deeply connected. The optimal decision is not "should we mark down?" but "should we mark down, or should we move the stock?"

Stage 6: End-of-season management

End-of-season (EOS) management covers the final decisions about inventory that will not sell at current prices: returns, final clearance, write-offs, and strategic consolidation.

Returns routing

Customer returns and inter-company returns need to go somewhere. The default in most retail operations is to return stock to the nearest store or back to the warehouse. Neither is optimal.

Returns and markdown intelligence routes returned stock to wherever it has the highest probability of selling. A returned size 7 does not go back to the store it came from (which might be overstocked). It goes to the store in the network with the highest demand for that size.

Consolidation

At end of season, many stores have fragments: a few units of various sizes that do not form sellable assortments. Consolidating these fragments into a small number of "clearance hub" stores creates sellable offerings from previously unsellable fragments.

This is what Replenify's CONSOLIDATE_REGIONAL algorithm handles in Pass 3 of the rebalancing engine. The bin-packing approach identifies which fragments can be combined into viable size runs and routes them to the stores best positioned to clear them.

Learning and feedback

The most neglected part of end-of-season management is the feedback loop. What worked? What did not? Which allocation decisions were right, and which created the problems that rebalancing had to fix?

This learning should feed directly into assortment planning for the next season. Which styles outperformed? Which stores' demand profiles shifted? Where did the size curves change?

Without structured learning, merchandise planners make the same mistakes season after season. With it, each season starts from a better baseline than the last.

Technology for merchandise planning

The technology landscape for merchandise planning ranges from purely manual to fully automated. Most mid-market retailers sit somewhere in the middle, with pockets of automation alongside significant manual effort.

The spreadsheet baseline

Spreadsheets remain the dominant planning tool for mid-market South African modern retail. A typical merchandise planner manages allocation tables, store grade matrices, and markdown schedules in Excel.

Spreadsheets are flexible, familiar, and free (relative to specialised software). They are also:

• Error-prone. Formula errors in allocation spreadsheets can misallocate thousands of units before anyone notices.
• Knowledge-dependent. The spreadsheet makes sense to the planner who built it. When that planner leaves, the replacement struggles to interpret the logic.
• Impossible to audit. "Why did Store 14 get 23 units of style 8842?" In a spreadsheet, the answer is buried in a chain of cell references. In a purpose-built system, the answer is a transparent decision trail.
• Disconnected. Allocation spreadsheets do not talk to rebalancing spreadsheets, which do not talk to markdown spreadsheets. Each function operates in isolation.

ERP-native functionality

ERPs like Posibolt provide transactional infrastructure: purchase orders, goods receipts, stock transfers, sales transactions. Some offer basic planning functionality (stock reports, reorder point alerts).

What ERPs typically do not provide:

• Optimised allocation across a multi-store network
• Distance-aware rebalancing recommendations
• Markdown timing and depth optimisation
• Closed-loop GP measurement
• Demand forecasting beyond simple moving averages

ERPs are essential as the system of record. They are not designed to be the system of intelligence.

Purpose-built intelligence platforms

This is where platforms like Replenify sit. They integrate with the ERP (through Replenify's canonical data model and Posibolt-native integration), consume the transactional data, and add the intelligence layer:

• Rebalancing: Deterministic engine with 3-pass architecture and 17-stage NAE pipeline
• Allocation: Per-store size curves, stockout bias correction, style-store affinity
• Pricing: Markdown timing and depth at store level, margin floor protection
• Replenishment: Lead-time-aware reorder optimization with safety stock intelligence
• Returns & Markdown: Returns routing to demand, network-wide markdown coordination
• Assortment Planning: Range architecture, store clustering, size/colour/style matrix optimization

The intelligence layer sits on top of the ERP. Retailers keep their existing operational systems and gain planning capabilities that spreadsheets cannot provide.

AI and machine learning in merchandise planning

The role of AI in merchandise planning is genuine but often overstated by vendors. Here is an honest assessment.

Where AI helps:

• Demand forecasting: ML models can identify non-obvious patterns in historical data and external signals
• Anomaly detection: Identifying unusual patterns (sudden demand spikes, data quality issues) that rules-based systems miss
• Natural language interaction: Asking questions about your inventory in plain language (this is what Reppi provides)
• Trend identification: Spotting emerging patterns across large datasets

Where deterministic logic is better:

• Allocation decisions: These need to be reproducible, auditable, and explainable. "The AI decided" is not acceptable when a buyer needs to explain why Store 14 got 23 units.
• Rebalancing execution: Same inputs must produce same outputs. The audit trail matters.
• Constraint satisfaction: Respecting OTB budgets, capacity limits, and margin floors requires deterministic enforcement, not probabilistic suggestions.

Replenify's approach -- "deterministic where it matters, AI-enhanced where it helps" -- reflects this reality. The core engines are deterministic and fully auditable. AI enhances the intelligence that feeds into those engines.

The shift from spreadsheets to intelligence

The transition from spreadsheet-based planning to an intelligence platform is not just a technology change. It is a shift in how the planning team works.

What changes

• From periodic to continuous. Instead of a weekly planning cycle, insights and recommendations flow continuously. The rebalancing engine can run daily.
• From isolated to connected. Allocation, rebalancing, pricing, and replenishment share a common data foundation. A decision in one area automatically informs the others.
• From assumption to measurement. Instead of "we think this allocation worked," you know it did -- or did not -- based on closed-loop GP measurement.
• From gut feel to data-informed intuition. Experienced planners do not become irrelevant. Their intuition becomes sharper because it is augmented by data they could not previously access.

What does not change

• Planners are still essential. Technology handles the computational heavy lifting. Planners handle the strategic judgement: which trends to follow, which risks to take, how to balance the art and science of fashion merchandising.
• ERP remains the system of record. Operational transactions continue to flow through the existing ERP. The intelligence platform reads from and recommends to the ERP; it does not replace it.
• Vendor relationships still matter. No algorithm replaces the buyer's relationship with suppliers, knowledge of production capabilities, or feel for emerging trends.

Getting started

The practical path from spreadsheets to intelligence:

1. Start with data. Ensure your ERP data is clean and accessible. This is the foundation everything else builds on.
2. Start with one module. Rebalancing is the natural starting point: it works with existing inventory, delivers measurable ROI quickly, and builds the data foundation for other modules.
3. Measure. Track the GP impact of every recommendation. This builds confidence in the system and creates the business case for expanding to other modules.
4. Expand. Add allocation, pricing, replenishment, returns management, and assortment planning as the data deepens and the team's confidence grows.

Merchandise planning is the highest-leverage activity in modern retail. Every other function -- operations, marketing, visual merchandising, store management -- operates within the constraints set by planning decisions. Getting those decisions right, supported by the right technology and the right data, is the difference between retailers that thrive and retailers that survive.

For a deeper look at specific aspects of merchandise planning, explore:

• How 3-Pass Rebalancing Works -- technical deep dive into in-season optimization
• The Stockout Bias Problem -- how allocation errors compound over time
• What 28,000 Inventory Snapshots Taught Us -- real data insights from the pilot
• Inventory Optimization: The Complete Guide -- the companion guide focused on optimization strategies and KPIs