There is a pattern in retail that has been proven again and again: the quality of a brand’s operations ultimately shows up in two numbers: full-price sell-through and inventory turnover days. The first reflects how strongly customers accept the merchandise. The second reflects how efficiently capital is being used. Neither number can be improved sustainably through promotions alone. Both are built up through the accuracy of many merchandise decisions over time.
The value of an AI merchandise decision system in footwear and apparel retail is not a vague promise of “efficiency improvement.” Its value lies in systematically reducing decision bias across six specific areas: pre-season assortment planning, allocation, replenishment, OTB, transfers, and in-season monitoring. When decision bias is reduced, full-price sell-through rises and inventory turnover days fall. The compounding effect of improving these two metrics is what truly improves operating quality.
What Does Operating Quality Really Mean?
In footwear and apparel, the term operating quality is used frequently, but most discussions stay at a high level. When we bring it down to merchandise management, operating quality can be broken into four measurable indicators:
Full-price sell-through: the share of seasonal merchandise sold at full price or with light markdowns. The higher this number, the more accurate the merchandise planning, allocation, and replenishment decisions are. It is one of the most direct financial reflections of merchandise decision quality.
Inventory turnover days: the average number of days from product receipt to sale. Faster turnover means less capital tied up in inventory and stronger resilience against inventory risk.
Size availability: the degree to which selling styles maintain a complete size range in stores. Size breaks are one of the most hidden forms of sales loss. When a customer comes in but cannot find their size, that revenue disappears silently.
Replenishment response cycle: the time from identifying a replenishment need to getting product into the store. The slower the response, the more sales window is lost. This is especially significant for styles with short four-to-six-week lifecycles.
An AI merchandise decision system affects the decision chain behind these four metrics. It does not create sales out of thin air. It reduces bias at each decision point, helping these metrics move in the right direction together.
1.Pre-Season Assortment Planning: How AI Reduces Guesswork
In footwear and apparel, merchandise planning is typically completed three to six months before the season. The core decisions at this stage are what to buy, how much to buy, and when to launch. The quality of these decisions directly determines the health of the inventory structure for the entire season.
Traditional planning relies heavily on buyer experience and trend intuition. This worked well when SKU counts were limited and channels were simple. But when a brand manages hundreds of SKUs, dozens of stores, and multiple online and offline channels, the boundaries of individual experience are quickly exceeded. Buyers may be able to judge the direction of hero products, but demand forecasting for mid- and long-tail SKUs is often closer to guessing.
What AI Does in Pre-Season Planning
AI does not replace buyers’ aesthetic judgment. It performs the calculations that buyers cannot reasonably do manually. These include:
Matching new styles with historical similar styles. When a buyer confirms a new style during the order meeting, the system can search the historical SKU library for comparable products based on attributes such as color family, silhouette, price band, and weight. It then extracts historical sales curves as a reference baseline and provides a recommended range for the first buy. This is not prophecy. It is a structured benchmark based on comparable data.
Category structure optimization. Based on sell-through and gross margin contribution over the past three to five seasons, the system can identify categories that are over-allocated and categories with clear demand gaps, helping buyers adjust both style breadth and buying depth.
Pre-season size curve recommendations. Based on actual store-level sales history, the system can generate differentiated size ratio recommendations for different regions and channel types before the season, instead of using one national parameter.
Key point: the value of AI in pre-season planning is not that it provides a final answer. It turns hidden experience into structured recommendations that can be reviewed. Buyer judgment remains central, but the quality of the evidence behind that judgment improves.
2.Allocation: From One Standard Ratio to Store-Profile-Driven Decisions
Allocation is one of the most underestimated profit levers in the merchandise planning chain. Moving products from the order plan into stores may look like an execution step, but it is a critical decision that shapes inventory health for the entire season. Many brands put significant effort into style selection, but when it comes to allocation, they still use one national size ratio across all stores.
The problem is that a unified ratio is essentially an average. It hides the real differences between store customer bases.
Consider a sneaker allocated by one national size curve: sizes 37 and 38 each account for 15%, sizes 39 and 40 each account for 30%, and size 41 accounts for 10%. After launch, stores in core East China trade areas sell out of sizes 37 and 38 first and remain out of stock for two weeks. Meanwhile, stores in Northeast China still hold large quantities of size 37 at the end of the season and eventually clear them at a discount.
The loss in both regions came from the same allocation decision, affecting roughly 18% of the style’s in-season gross margin. The product was not necessarily a poor seller. It was allocated to the wrong places.
Traditional allocation logic: one national size ratio → regional average adjustment → unified distribution. Allocation accuracy stops at the regional level, and individual store differences are averaged away.
AI allocation logic: multidimensional size curves by store, category, and season → calibration based on actual sales → store-level differentiated recommendations. Allocation parameters are calculated independently for each store.
The financial meaning of more precise allocation is clear: it reduces markdown loss from excess non-core sizes while reducing opportunity loss from core-size stockouts. In footwear and fall/winter apparel categories, where size structures are complex, the combined impact of these two types of loss on in-season gross margin can be significant.
3.Replenishment: From Replenishing After Stockout to Triggering Alerts Earlier
Replenishment is the first correction mechanism after allocation bias appears. It is also one of the most time-consuming daily activities for merchandise teams. But the quality of replenishment decisions depends heavily on timing, not just quantity. For fast-moving footwear and apparel styles, the peak sell-through period is usually concentrated in the first three to four weeks after launch. If the replenishment process takes one to two weeks, then even a correct replenishment decision may arrive after the sales peak has passed.
Weeks of supply, or WOS, is a core early-warning metric for store inventory health. The formula is: WOS = current inventory ÷ average daily sales. The number indicates how many weeks of normal sales the current inventory can support at the current sales rate. Low WOS signals an upcoming stockout and should trigger replenishment. Persistently high WOS signals rising overstock risk and should trigger transfer or promotional action earlier.
Predictive WOS Triggering
AI uses forecasted future sales together with actual WOS conditions to trigger replenishment before a stockout occurs, rather than waiting until the product is already out of stock. Recommended replenishment quantities also need to be precise. Simply using historical monthly average sales as a replenishment baseline ignores changes in current sales velocity, holiday effects, promotion calendars, and other dynamic factors. An AI system combines recent sales velocity, forecasted sales, in-transit inventory, and available store capacity to recommend quantities, reducing the cycle of over-replenishing and creating new overstock.
7thonline’s replenishment module is built around daily dynamic WOS calculations. It continuously monitors inventory health at the store × SKU level. For styles with continuously accelerating sell-through, the system actively flags the reorder window, including how much time remains before the supplier’s latest reorder cutoff. This helps buying teams act within the effective response window and avoid missing supply chain opportunities because of decision delays.
For one domestic fashion brand using the system, the merchandise control team reduced replenishment response time from an average of 7 to 15 days to 1 to 2 business days. The team’s daily work shifted from actively searching for problems to reviewing system recommendations and confirming execution.
4.Dynamic OTB Management: Keeping Budget Aligned with Sales Rhythm
Open-to-buy, or OTB, is the budget control tool for merchandise planning. Most brands lock OTB after the pre-season order meeting and do not adjust it during the season. This logic made sense when markets were stable and sales structures were predictable. Today, the problem is that sales structure changes continuously during the season, while budget allocation remains static.
The result is that fast-selling categories have no remaining budget for chase orders, while slow-moving categories continue to tie up capital. OTB, which should be a resource allocation tool, becomes a rigid rule that limits flexibility.
The problem with OTB is not the total budget. It is that budget allocation does not update in sync with sales performance. This is one of the most common and overlooked forms of resource waste in in-season merchandise management.
AI-driven dynamic OTB uses a rolling mechanism by category. Every day, it recalculates remaining available budget by category based on actual sales progress, remaining selling period, and current inventory levels, then pushes adjustment recommendations to relevant decision-makers. When a category performs above plan for two consecutive weeks, the system releases chase-order budget space for that category. When a category continues to sell below expectations, the system reduces further investment earlier.
The mechanism itself is not complicated. But executing it requires real-time multi-category sales data and automatic recalculation, which is exactly where manual operation becomes a bottleneck. A weekly in-season meeting can only discuss a limited number of categories. A system can monitor all categories at the same time.
5.Transfers: Replacing Gut Feel with Return-Based Calculation
Transfers are the most execution-heavy and often the most delayed of the three core merchandise actions: allocation, replenishment, and transfer. The reason is simple. Transfer costs, including logistics, management friction, and store cooperation, are certain. But the benefit, such as incremental sales at the receiving store and reduced markdown loss at the sending store, is estimated. When the benefit is unclear, teams are likely to delay or avoid the transfer.
The core job of an AI transfer system is to make that estimate clearer:
Scan network-wide inventory and identify supply-demand mismatches such as Store A being overstocked while Store B is short on the same style.
Estimate expected incremental sell-through after transfer based on the receiving store’s remaining selling window and recent sales velocity.
Combine logistics cost to calculate whether the net return is positive and rank the priority.
Output a transfer recommendation list with expected return estimates for merchandise teams to review and execute.
The shift from “it feels like we should transfer this” to “the system shows a positive net return and recommends execution” changes decision confidence. When confidence improves, execution rates rise.
Frequent transfers usually indicate that something went wrong at the allocation stage. Transfers should be exceptions, not a routine task every season. Once AI improves allocation precision, transfer demand naturally declines. That is the real direction of optimization.
6.In-Season Monitoring: Making Product Status Visible in Real Time
After pre-season planning is completed and products are launched, most brands rely on weekly reports and monthly reviews for monitoring. The issue is that sales divergence for a single footwear or apparel style often appears in the second or third week after launch. Which stores are accelerating, which sizes are shifting, and which regions are showing structural slow movement can all become visible early. By the time these issues are discussed in a monthly review, the intervention window may already be half closed.
An AI product health monitoring system scans all active SKUs every day. It combines multiple signals, including changes in sales velocity, inventory-to-sales ratio trends, and regional distribution breadth, and classifies product status in real time as accelerating sell-through, slowing momentum, structural shift, and other types. When exceptions are triggered, the system pushes alerts proactively. Merchandise teams no longer need to dig through reports every day to find problems. The system brings the products that need attention directly to them.
This changes more than efficiency. It changes how merchandise teams work: from “we go looking for problems” to “problems come to us.” Limited human judgment can then be focused on decisions that truly require human intervention.
The value of an AI merchandise decision system in footwear and apparel retail is not a broad promise of efficiency improvement. It lies in reducing decision bias across six concrete areas: the quality of pre-season planning references, the structural precision of allocation, the timing of replenishment triggers, dynamic OTB adjustment, visibility into transfer returns, and real-time in-season sensing.
When decision bias is reduced, full-price sell-through improves and inventory turnover days decrease. The compounding effect of improving these two numbers is what truly improves operating quality.
7thonline focuses on merchandise management for footwear and apparel retail. It has served global and domestic brands including Patagonia, Calvin Klein, Canada Goose, Under Armour, and Ellassay across scenarios such as precise allocation, replenishment, and transfer; dynamic in-season OTB management; and store-level size modeling. If you would like to explore implementation options for specific scenarios, please contact us for a dedicated diagnostic assessment.
