Pick-face replenishment moves eligible stock from reserve storage into the locations used for order picking. Its purpose is to make the required item available where the next pick will actually happen, before demand exhausts that location.
The warehouse has the goods. The picker reaches the shelf. The shelf is empty. Somewhere above the aisle, or across the building, a reserve pallet contains exactly what the customer ordered. The stock exists, but the order cannot use it yet.
This is a different failure from buying too little inventory. It is a failure to bring existing inventory into the right working position at the right time. Adding more reserve stock may make the warehouse fuller without making the empty picking location any more useful.
Oracle’s warehouse replenishment documentation describes this reserve-to-picking movement and distinguishes several ways to trigger it. This guide develops the operational question behind those functions: how does a warehouse keep picking supplied without filling every accessible space or flooding the aisles with replenishment work?
This is Article 121 in the extended How Logistics Works collection. Putaway and Slotting explains where stock should live; Order Picking explains how it is selected. Here, the focus is the repeated internal supply movement connecting the two—not purchasing, supplier replenishment or the wider inventory policy.
One product can have two very different kinds of availability
Reserve storage and forward picking serve different purposes. Reserve storage can hold larger quantities in a form suited to storage and bulk movement. A forward-pick location provides smaller quantities in a position and presentation suited to repeated selection. The same product may exist in both places, but it is not equally accessible to the same task.
A full pallet high in reserve storage may require equipment and an authorised operator before it can support a small each-pick order. A case in the picking location may be immediately usable. Counting the two together is useful for total inventory, but insufficient for deciding whether the next hour of picking can proceed.
Consider three questions separately. Does enough stock exist? Is that stock eligible for these orders? Can it reach the picking location before those orders need it? A replenishment problem can sit in any of the three, and each calls for a different response.
The pick face is a working buffer, not a miniature warehouse
A large pick face can reduce replenishment frequency. It also occupies accessible space that another product could use. A small pick face can improve density and shorten travel, but may need frequent replenishment. Neither arrangement is universally efficient.
The useful size depends on demand, replenishment response time, packaging increments and physical capacity. A fast-selling product replenished promptly may need less forward stock than a slower-selling product whose reserve movement is difficult. Demand alone does not determine the answer.
The design therefore has two linked decisions: how much working stock to position and how reliably to renew it. Increasing the first can hide weakness in the second. Improving the second can sometimes release valuable space without reducing service. The Logistics Buffers guide provides the wider explanation of this trade-off.
Follow the transfer, not just the replenishment request
The internal journey begins when the system or operator identifies a need. Eligible reserve stock is selected, movement work is created, equipment and labour become available, the stock is retrieved, transported and placed in the destination. The destination quantity and status are then confirmed.
A request is not a completed transfer. A task assigned to a forklift is not stock in the pick face. A pallet standing beside the destination may still require breakdown, counting or a recorded put before the picking process can use it.
Microsoft’s min/max replenishment example represents replenishment work with both pick and put steps. That software example reflects an important physical distinction: removing inventory from reserve and making it available at its destination are separate events. The operation needs evidence of both.
A trigger should answer a specific kind of shortage
Microsoft’s replenishment overview distinguishes minimum/maximum replenishment, wave-demand replenishment, load-demand replenishment and immediate replenishment. These are documented application functions, not one universal vocabulary that every warehouse system must use.
Operationally, the distinction is between replenishing because a location has fallen below a defined level, replenishing because known work needs more stock, and reacting because a shortage has become immediately relevant. Each trigger sees a different part of the future.
A threshold can protect recurring demand before individual orders are released. A demand-based trigger can prepare an unusual concentration of work. A reactive trigger can recover from an unexpected gap. A warehouse may need a combination, provided the processes recognise one another’s open work rather than creating duplicate transfers.
Minimum and maximum are decisions, not magic settings
The minimum should leave enough time to act before stock becomes unusable for the intended picking workload. The maximum should fit the location and the operating design. Copying yesterday’s settings into a changed order profile can preserve the appearance of control while weakening the actual response.
A minimum of twenty units means little without a consumption rate. Twenty units may support an entire shift for one product and a few minutes for another. It also means little without knowing how long replenishment takes once requested.
Choose the review interval as carefully as the threshold. A location that falls below its minimum just after a periodic check may wait almost a complete review cycle before work is even created. The practical response time begins with recognition of the need, not merely with the eventual vehicle journey.
Known demand can reveal a shortage before the shelf looks empty
A pick face can appear comfortably stocked while the next released group of orders requires more than it holds. Looking only at the shelf misses the committed demand approaching it. Demand-based replenishment looks ahead to the work the location is expected to support.
That does not mean every future order should immediately become a replenishment task. Future work may change, stock may already be moving and accessible space may be limited. The trigger needs a defined planning horizon and a clear account of existing commitments.
Oracle’s replenishment-with-picking-wave workflow illustrates how order need, current stock, allocated quantities and in-transit replenishment can interact. Its particular availability calculation is application-specific. The transferable lesson is to represent these states separately instead of counting every quantity as immediately pickable.
Compare the stock clock with the replenishment clock
A simple diagnostic model estimates time to depletion by dividing the usable physical quantity in the pick face by the expected consumption rate. It assumes a reasonably steady rate over the interval and no earlier incoming transfer. It is a starting calculation, not a service guarantee.
Illustrative time to depletion = eligible physical quantity at the pick face ÷ expected picking consumption per unit of time.
Compare that estimate with the complete replenishment response: recognition, queue, travel, retrieval, transfer, placement and confirmation. A shelf with forty minutes of supply is already exposed when replenishment usually takes fifty minutes.
Keep the arithmetic consistent. When consumption includes all planned picks, do not subtract those same picks from the starting quantity and then count them again in the depletion rate. Reservation, physical stock and expected demand can be represented in several valid ways; mixing the conventions creates a false shortage or false assurance.
A worked example: enough stock, but nine minutes too late
The following example is hypothetical. At 09:00, a picking location contains seventy-two eligible units. Demand consumes two units per minute. Reserve storage contains several hundred more units, so total inventory is not the immediate constraint.
Under the simple steady-rate assumption, the pick face empties after thirty-six minutes. A replenishment completed in thirty minutes arrives with twelve units still available. A replenishment completed in forty-five minutes arrives nine minutes after the location has emptied.
The average replenishment time can therefore look acceptable while slower responses still interrupt picking. The relevant question is not only the mean. Which delays occur during the actual busy period, and how much stock remains when those delays occur?
Triggering at ninety units would provide forty-five minutes of coverage at the same assumed rate. It would not protect against every surge, delay or eligibility change. The example shows how to connect quantity to response time; it does not prescribe a ninety-unit minimum for a real warehouse.
The team has several possible interventions. It could detect the need earlier, shorten the replenishment queue, improve reserve accessibility, adjust the forward allocation or coordinate the order release. The right repair depends on which part of the clock is consuming the margin. More stock is only one option.
Open replenishment is future capacity, not current stock
Suppose the system has created work to move four cases into the picking location. A second trigger runs before the first transfer finishes. Without awareness of open work, it may create another four-case movement and overload the destination.
Counting expected incoming stock can prevent that duplication. But treating it as physically present can release picks that arrive before the transfer. The operation needs both views: what is available now and what is expected to become available, with an owner and an estimated completion.
When a transfer fails, its promised quantity must not remain indefinitely in the planning picture. A cancelled or blocked task changes future availability. That change should reach whichever process is deciding whether the next work can safely be released.
Capacity is physical occupancy, not an optimistic spreadsheet balance
A replenishment destination needs enough room for the arriving stock. Free capacity depends on its dimensions, weight limits, packaging and mixing rules—not merely on the numerical difference between a maximum quantity and a recorded balance.
Case multiples can complicate the decision. A location may have space for eighteen more units while replenishment normally moves twelve-unit cases. Moving two cases would exceed the defined limit. Moving one leaves unused capacity but remains feasible. Breaking a case may be possible only under a suitable process.
Microsoft’s replenishment-strategy documentation distinguishes covering wave demand from replenishing selected locations to configured maximum capacity. The latter can reduce repeated work, but it depends on valid stocking limits and sufficient source inventory. Filling more is a strategy with conditions, not an automatic improvement.
Eligibility follows the goods into the pick face
A reserve pallet is not usable merely because its product code matches. It may be held, allocated to another requirement or incompatible with an order’s lot, expiry, ownership or condition requirement. Replenishment should not erase those distinctions.
The destination also matters. Moving two lots into one location may be allowed in one process and unsuitable in another. A mixed location that the picking method cannot distinguish reliably can turn a replenishment success into a selection error.
Keep specialist rules with their qualified owners. This guide does not decide product disposition or set expiry rules. Its narrower point is that moving stock internally must preserve the eligibility conditions that determined whether it could be used at all.
Replenishment and picking share space as well as inventory
The two processes can compete for an aisle, a workstation or an equipment route. An emergency replenishment may restore one product while interrupting several pickers. A large batch of replenishment work may fill locations efficiently and make the area temporarily difficult to use.
Coordinate the activity through the site’s approved traffic and safety arrangements. Possible operating designs include planned replenishment windows, physically separated access or smaller movements that fit the work pattern. Their suitability depends on the facility; production pressure is not a reason to bypass safe separation.
Measure the combined effect. A replenishment operator completing more transfers per hour is not necessarily improving fulfilment when the transfers create larger picking delays. The local task rate needs to be read beside the supported picking result.
An emergency queue can become the normal process
A reactive replenishment is useful when reality changes unexpectedly. When the same product repeatedly needs urgent attention, the reaction may be compensating for a weak threshold, stale demand assumption, unreliable source location or insufficient transfer capacity.
Do not merely celebrate how quickly emergencies are cleared. Follow recurrence by product, time band and cause. A location that fails every afternoon may need a different pre-peak preparation rather than a more persuasive urgency label.
Also protect genuinely urgent work from routine overclassification. When every replenishment is critical, the priority queue stops communicating anything. The remaining stock horizon and downstream consequence provide a more useful basis for triage than who asks most forcefully.
A faster picker can expose a previously hidden replenishment limit
A new picking method reduces travel and raises consumption at the forward location. Nothing about the replenishment settings changes. The warehouse then experiences more empty-bin interruptions and may conclude that the new picking system is unreliable.
The local improvement may be real. It has changed the demand imposed on the internal supply process. Recalculate the depletion horizon and examine whether replenishment can sustain it. Capacity is a relationship between connected steps.
The same reasoning applies when robots bring inventory to people. Physical walking may fall while the replenishment or decanting operation becomes more important. Automation changes the location of work; it does not make the supporting stock flow unnecessary.
Confirmation is where inventory truth is restored
A reliable transfer records the source, destination, item and actual quantity moved, along with relevant status or tracking information. It does not simply increase the destination balance and hope that reserve will reconcile later.
Partial transfers need explicit treatment. If three of four requested cases are moved, the remaining case is still an open requirement or a deliberately cancelled quantity. Marking the whole task complete can release picking work against stock that never arrived.
A discrepancy should preserve a known holding state and a next owner. The Logistics Reconciliation guide explains how to resolve conflicting records without making the screens agree through unsupported adjustments.
Measure the stockout that happens despite stock being present
A particularly useful diagnostic is the occurrence of pick-face shortages while eligible reserve stock exists. It separates an internal positioning failure from a true inventory shortage. Define the population and observation method carefully so unrecorded reserve holds are not misclassified as available stock.
Pair that measure with time from replenishment need to usable destination receipt, emergency share, repeated transfers, short transfers and picking delay. Follow the slow tail as well as the average. A small number of very late movements can disrupt important orders even when the overall mean improves.
For productivity, include the work created elsewhere. Fewer reserve trips may be attractive, but not when excessive forward stock causes more relocations or obstructed access. The outcome to protect is accurate, continuous fulfilment with an appropriate resource burden.
Improve one recurring pattern before retuning the entire warehouse
Select a product-location combination with repeated interruption and trustworthy records. Reconstruct demand and the complete transfer timeline. Establish whether the issue is late detection, unavailable source stock, queueing, physical travel, destination fit or confirmation.
Test a bounded change against a stated prediction. An earlier trigger should create more response margin; a better source location should shorten a particular travel or retrieval step. Observe the mechanism, the picking outcome and any extra space or handling burden.
Expand only when the result supports expansion. A successful morning test does not establish peak-period performance. Preserve the conditions under which the change worked, update the relevant settings and teach the exception path to the next shift. This connects replenishment to the collection’s Learning Loops.
Questions worth asking at the empty shelf
Is this a procurement problem? Not necessarily. First determine whether eligible reserve stock exists. A warehouse can own enough inventory and still fail to position it for the current task.
Should every pick face be full? No. Full locations consume accessible space and may create unnecessary handling. The appropriate quantity depends on demand, transfer response, physical capacity and the role of the location.
Can the system count stock already being replenished? It may need to count that quantity for planning and duplicate prevention. It should still distinguish expected stock from stock physically confirmed at the picking location.
What should improve first? Find the interval where usable supply is lost. A larger threshold will not solve a release hold; faster equipment will not correct the wrong case conversion. The next action should follow the mechanism actually observed.
Evidence and use of the examples
The primary implementation references are Oracle’s replenishment overview and picking-wave workflow, together with Microsoft’s replenishment overview and strategy documentation. They demonstrate documented software behaviours, not universal configuration instructions or independent performance benchmarks.
The clock model and numerical scenario are original illustrations. Their assumptions are stated so the reasoning can be checked. Real thresholds require local demand, variability, capacity and safety evidence. Editorial evidence review: 5 September 2026.
Return to the order that needs the stock
Return to How Logistics Works for the complete movement-and-receipt system. Replenishment adds a precise lesson: available inventory must also be available to the next process, in the right place and soon enough to be useful.
The successful replenishment is not the pallet that left reserve. It is the stock that reached the picking location, retained its identity and eligibility, and became usable before the next order ran out of time.