# End-to-end supply-chain visibility for allocation and replenishment
A cotton crewneck leaves a factory in Tirupur, India, on a Tuesday. Six weeks later it should be folded on a table in a Chicago flagship. Between those two moments sit a purchase order, an ocean container, a customs hold, a distribution center, and a decision: which of your 200 stores gets those 5,000 units, and how many.
Get that decision wrong and you create the two most expensive words in retail: stockout and markdown. Get it right, using data you already have, and you sell more full-price units from the same inventory.
This lesson traces that crewneck from factory PO to store shelf, and shows how lead-time and in-transit data feed an allocation model that rebalances inventory before stockouts hit.
Let's define the terms as we go.
A purchase order (PO) is your commitment to a factory: style, quantity, price, ship date. Once cut, that quantity is largely fixed. This is why fashion is a forecasting business, not a reordering business, for a lot of its assortment.
Here is the typical path and where data gets captured:
Each step produces a timestamp. String those timestamps together and you get lead time: the elapsed days from PO to available-to-sell. This is the single most important number in this lesson.
Most brands plan against a *standard* lead time. Say the planning system assumes 42 days factory-to-DC. Reality varies wildly: a monsoon delay, a port strike, a mislabeled container.
If you allocate against the assumed date and the goods land 12 days late, your fast-selling stores go dark right through a peak weekend. If they land early and you over-shipped a slow store, you are now shuffling boxes between locations, which is expensive and slow.
The fix is to plan against actual, live in-transit data, not a static assumption. Modern GS1 EPCIS event standards (a common language for "what happened, where, when" across trading partners) make this feasible even across many vendors. See the GS1 EPCIS standard overview for how these events are structured.
The mindset shift: treat lead time as a distribution, not a single number.
Allocation answers one question: given X units arriving on day D, how many go to each location?
You need four inputs, all data you likely already have:
1. Demand rate per store: recent rate of sale, ideally by size. A store selling 8 units a week of a size Medium needs more Mediums.
2. On-hand + on-order: current inventory plus anything already inboundinboundA strategy that attracts prospects organically via valuable content (blog, SEO, social) rather than interrupting them.View full definition → to that store.
3. Target weeks of supply (WOS): how many weeks of cover you want on the shelf. A core basic might target 6 to 8 weeks; a trend item, 2 to 3.
4. Lead-time-to-store: DC-to-store transit, which differs by region.
The core formula for what a store *needs* is straightforward:
# Units a store needs to reach its target cover
need = (target_wos * weekly_rate_of_sale) - on_hand - on_order
# Never negative, and respect the size curve minimums
need = max(need, 0)
# When arriving supply is scarce, allocate proportionally to need
store_alloc = round(available_units * (need / total_need_all_stores))That last line is the important one. When you cannot fully cover every store (the normal case), you split scarce units in proportion to demonstrated need, not evenly. Equal splits are how you end up with 3 units in a store that sells 20 a week and 3 in a store that sells 1.
Here is where in-transit data earns its keep.
You do not have to wait until goods hit the DC to decide. If your EPCIS or carrier feed says the container clears customs in four days, you can pre-plan the allocation now and cut the store shipments the hour goods are received. That can shave days off shelf time.
Better still, you can rebalance before the stockout. Say your live rate-of-sale data shows a color running hot in the Northeast and cold in the Southwest. Two levers:
The discipline is comparing days of cover remaining against days until replenishment lands. If cover is 6 days and the next truck is 9 days out, that store will stock out for 3 days. That gap is your alert. It is far more actionable than a static "low stock" flag.
Allocation is not just store-to-store. Most brands run an omnichannelomnichannelAn integrated approach connecting all customer touchpoints (physical, digital, mobile) into a seamless experience, with shared data and consistent context across channels.View full definition → model (selling across physical and digital channels from shared or separate inventory pools).
The tension: e-commerce demand is national and can spike overnight; store demand is local and steadier. If you dump all ATS into stores, your website goes out of stock on a bestseller and you lose the highest-margin, no-markdown sale. If you hoard everything for e-commerce, stores look empty and understocked.
A practical rule many brands use: hold a central buffer at the DC that can serve e-commerce *and* backfill any store that runs hot, rather than committing every unit to a physical location on day one. Live lead-time data tells you how big that buffer needs to be. Short, reliable lead times mean a smaller buffer; long, volatile ones mean a bigger cushion.
Knowledge check
1. Why is fashion described as "a forecasting business, not a reordering business" for much of its assortment?
2. A unit of inventory becomes relevant to an allocation decision at which point in the journey?
3. What is the core reason end-to-end visibility (lead-time and in-transit data) matters for allocation and replenishment?
4. Select ALL correct answers about lead time and the timestamps captured along the PO journey.
Select all the correct answers.
5. Select ALL correct answers describing the risks of getting the allocation decision wrong.
Select all the correct answers.
Visibility only matters if it changes a decision. Watch these three:
1. Lead-time variance. Not the average, the spread. If a lane's transit swings between 35 and 60 days, your safety stock has to cover the bad case. Reducing *variance* is often worth more than reducing average speed.
2. Sell-through rate. Units sold divided by units received, over a period. High and early sell-through on limited stock is your signal to chase transfers or expedite; low sell-through is your signal to stop feeding that store and start planning markdown.
3. In-stock rate by size. A style can look "in stock" while missing the two sizes that sell most. Tracking availability at the size level, not just the style level, is where a lot of hidden lost sales live. Being out of Medium is being out of the sale.
The Tirupur crewneck: 5,000 units, cleared customs, available to sell Friday.
If total network need is 4,000 units and you have 5,000, everyone gets covered and 1,000 stays in the DC buffer. If total need is 8,000, you allocate proportionally: Store A gets its share of the shortfall, Store B gets its (smaller) share, and no single hot store starves a slow one of its trickle.
The visibility part: because you saw the customs clearance Wednesday, the allocation was already cut and shipments left the DC Friday afternoon, not the following Tuesday.