By Elie Dufeu, CTO & Co-Founder, Metreecs. Published 1 July 2026.
Every fashion season ends with the same paradox: some stores are running out of the styles that sell, while others are sitting on stock they cannot move. The gap between them is not a buying problem. It is a distribution problem, and in-season stock transfers are the mechanism to close it.
Inventory rebalancing, also called inter-store transfers or store-to-store stock movements, has historically been treated as a last resort: evidence of poor initial planning. That view is wrong, and it costs brands margin every season.
Key Takeaways
- Even accurate initial allocation produces imbalances mid-season: demand shifts faster than any pre-season forecast can anticipate
- Size-level granularity matters more than product-level totals. A store may look balanced in units while missing its top two selling sizes entirely
- Every transfer decision requires a net benefit calculation: margin recovered minus transport cost. Not every imbalance justifies a move
- On a flat size curve, a 1,000-unit buy can net roughly 11% less gross profit than the same buy allocated on the correct curve, largely from broken-size markdowns that rebalancing helps avoid
- AI-driven allocation and rebalancing has been documented to cut lost sales from stockouts by up to 90% in fashion retail deployments, in part by moving stock to where it actually sells
See how Metreecs handles store-level demand signals and transfer recommendations.
Why rebalancing has a bad reputation (and why that needs to change)
For decades, merchandising teams were taught that inter-store transfers are a sign of failure. The logic was simple: if you planned well, you would not need to move stock around mid-season. Put energy into the initial forecast, not the cleanup.
That logic held in an era of slower consumer behavior, fewer channels, and simpler assortments. It does not hold today.
A multi-store fashion brand selling several hundred active products in multiple colors and sizes per style is managing tens of thousands of store-product-size combinations. Each of those combinations has its own local demand curve. A style that sells quickly in one city can move at half that rate in another. A size that sells out at a flagship in week two may still have units sitting untouched in a secondary market in week five.
No initial allocation, however accurate, anticipates all of that. Demand variability at the store-size-color level is simply too high.
The consequence of refusing to rebalance is predictable: one store marks down to clear stock it could have moved elsewhere, while another discounts to clear a size it could have transferred out weeks earlier. Both markdowns are avoidable.
A pattern that shows up repeatedly in post-season reviews: a meaningful share of marked-down units were sitting in a surplus position in at least one other store, at the same time a different store was out of stock on that exact size and color. The transfers that could have prevented those markdowns were never triggered, because no one had time to calculate them manually across a large store network.
When to rebalance: the signals that matter
Not every stock imbalance justifies a transfer. The decision requires reading three signals simultaneously:
1. Sell-through rate divergence between stores
When one store is selling through a given size-color variant much faster than another store carrying the same product, that gap is a transfer signal. The sell-through velocity difference tells you where demand is and where it is not. The question is whether closing the gap is worth the logistics cost.
2. Time remaining in the selling window
A product with six weeks left before the markdown date has time to justify a transfer. The same product with 12 days left probably does not. Rebalancing decisions are time-sensitive: the earlier in the season you act, the more selling window remains to recover the transfer cost.
3. Net benefit calculation
This is the step most manual processes skip, and it is where most of the value sits. The transfer makes sense only when the margin recovered on transferred units exceeds the cost of transfer (picking, packing, transport, handling).
As an illustrative example: a product retailing at 150 euros with a 50% margin recovers 75 euros per unit sold at full price. Transferring four units at a total logistics cost of 30 euros produces a net benefit of 270 euros. Transferring a single unit of a 40-euro product at 15 euros logistics cost produces a net benefit of 5 euros, barely worth the operational overhead.
Scaling this calculation across hundreds of products and dozens of stores is impractical manually. It is exactly what AI-driven demand planning systems do continuously.
The size run problem: why unit totals mislead you
In fashion, aggregate unit counts are a misleading metric for rebalancing decisions. A store showing a healthy total unit count on a given style may hold most of those units in the largest and smallest sizes, while every size in between has sold out. That store has no effective inventory of the product for most of its customers.
The real unit of analysis in fashion rebalancing is the size-color combination, not the product as a whole.
Consider a typical scenario: a brand launches a style in several colorways across dozens of stores. After a few weeks, one location has sold out of its best-selling sizes in one colorway but still holds units in sizes that rarely move locally. A nearby store, serving a different customer base, is moving slowly on the exact sizes and colorway the first store needs.
A transfer between the two closes a real demand gap. A report showing only total units per store would never flag the problem.
This is one reason AI-driven inventory optimization has been linked to meaningfully fewer stockouts: the system evaluates demand at size-color granularity, not at the product level, and surfaces hidden imbalances that aggregate reporting masks. Learn more about AI-driven inventory optimization.
Rebalancing vs. replenishment: choosing the right lever
Before deciding to rebalance, rule out the faster alternative: replenishment from the distribution center.
If the distribution center still holds safety stock of the relevant product and size, a replenishment order is almost always cheaper and faster than a store-to-store transfer. Replenishment taps a single source. Store-to-store transfers involve coordinating two stores, multiple logistics routes, and potential disruption to the originating store's service level.
The rebalancing decision becomes relevant when the DC is out of stock on the relevant product and size, lead time from the DC exceeds the remaining selling window, or a specific store has clear excess at the same time as another store has a clear shortage. In practice, most mid-season rebalancing decisions arise from a combination of all three.
The demand planning layer matters here too: when forecast accuracy at the store-product level is high, the safety stock calculation behind it was sound. The gap being rebalanced is genuine demand variability, not a forecasting failure.
What AI does differently in rebalancing
The manual approach to rebalancing runs on spreadsheets and intuition. A merchandiser exports sell-through by store, scans for obvious outliers, and decides by judgment which transfers to recommend. In a large multi-store, multi-product environment, that process takes days per cycle and still misses most of the opportunity.
AI-driven rebalancing works differently on four dimensions:
Continuous monitoring instead of weekly snapshots. The system evaluates sell-through divergence at the store-product-size level daily, not weekly. Imbalances surface in hours rather than after a merchandiser has time to run a report.
Net benefit calculation at scale. For every potential transfer, the system calculates the expected margin recovery (probability of full-price sale times margin per unit) against the transfer cost. Transfers below a minimum net benefit threshold are filtered out automatically.
Route consolidation. Instead of recommending individual one-off transfers between stores, the system consolidates routes and minimizes truck rolls, cutting the logistics cost of the rebalancing program.
Broken size run completion. The system specifically identifies stores where a partial size run is suppressing demand, and prioritizes transfers that restore a complete, sellable size run.
Teams that move from weekly manual transfer planning to AI-assisted review commonly report going from a full day per week building recommendations to reviewing and approving a pre-calculated list in well under an hour, with the reclaimed time going toward decisions that still require judgment: which new collection to prioritize, how to handle a launch in a new market.
AI agents for retail are what enable this shift from reactive to continuous. The monitoring, the calculation, and the route optimization run automatically. The merchandiser acts on exceptions, not noise.
How to structure a rebalancing process that works
Whether you are running rebalancing manually or with AI support, the underlying process logic is the same:
Step 1: Set a monitoring cadence. Review sell-through divergence at least weekly. In a fast-moving fashion season, twice weekly is better. The earlier you catch an imbalance, the more selling window remains.
Step 2: Define a minimum net benefit threshold before the season starts. This prevents the team from spending more in logistics than the margin it recovers, and eliminates the noise of marginal transfer opportunities.
Step 3: Protect the originating store's service level. A transfer should not leave the sending store below its minimum presentation depth for the style. Calculate the transfer quantity against the originating store's current sell-through rate and days of selling window remaining.
Step 4: Track actual vs. expected benefit. After each rebalancing cycle, compare the expected margin recovery against the actual sell-through on transferred units. Over time, this builds the dataset that refines the net benefit model and improves future transfer decisions.
This feedback loop, tracking actual outcomes against the model's predictions, is what makes the system smarter over time and what narrows the gap between a rough forecast and a genuinely reliable one at the store-product level.
FAQ
What is the difference between inventory rebalancing and replenishment?
Replenishment restocks a store from a central source (distribution center or warehouse). Rebalancing moves stock from one store to another. Replenishment is generally faster and cheaper when the DC has inventory available. Rebalancing is the right lever when the DC is empty or the lead time exceeds the selling window.
How often should a fashion brand rebalance inventory?
For fast-moving items in a short fashion season, a weekly review cycle is the minimum. For slower-moving styles with longer lifecycles, every two weeks may be sufficient. The key is not the calendar frequency but the trigger: rebalancing should happen when the sell-through divergence crosses a meaningful threshold, not on a fixed schedule.
Does rebalancing make sense for brands with fewer than 10 stores?
Yes, though the math changes. With fewer locations, each transfer involves a larger percentage of total inventory for the relevant product. The net benefit calculation still applies: calculate the expected margin recovery against the transport cost before executing.
What causes inventory imbalances even when the initial allocation was accurate?
Demand changes. A promotion in one city, an unexpected competitor closure, a weather event, a viral moment on social media: any of these can shift sell-through velocity at the store level within days. No initial allocation survives contact with real demand variability at this granularity. Rebalancing is the mechanism for adapting to what actually happened, not what was forecast.
How does rebalancing affect days inventory outstanding (DIO)?
Done well, rebalancing reduces DIO by converting slow-moving stock at one location into full-price sales at another. Stock that would have ended the season as markdown inventory at one store sells at full price at another after the transfer, so the working capital tied up in that stock converts to realized revenue faster.
Is inventory rebalancing a sign of bad planning?
No. It is a sign of an operating environment with genuine demand variability. Even a strong forecast at the store-product level leaves some demand unaccounted for, and in a network of dozens of stores across hundreds of products, that residual translates to real store-level adjustments needed mid-season. The brands that treat rebalancing as a planning failure stop acting on those signals and pay for it in end-of-season markdowns.
Conclusion
Inventory rebalancing is not the last resort of a badly planned season. It is the operational discipline that converts avoidable markdowns into full-price sales by moving stock from where it is sitting to where it will sell.
The discipline works at any scale. What changes with scale is the data requirement. Manual processes stop being viable above a certain store count. AI-driven monitoring, net benefit calculation, and route optimization are what make continuous rebalancing practical at larger networks.
If your end-of-season analysis regularly shows units marked down in one store while another was out of stock on the same product, you are not facing a buying problem. Book a demo to see how Metreecs surfaces those gaps in real time, before the markdown window closes.
Sources
Our articles
Own your operations with AI Intelligence
.svg)