Replenishment Planning Demystified: Techniques for a Lean Inventory

What is replenishment planning, in one paragraph?

Replenishment planning is the process of deciding when to reorder a SKU, how much to reorder, and from which supplier or warehouse, so that available-to-sell stock matches forward demand without overshooting. In an apparel context, it is the operational layer that sits between your sales forecast and your purchase orders. It governs core carryover programs, basics, replenishment-eligible wholesale styles, and any DTC SKU that you intend to keep on the floor for more than one season. Done well, it produces a short, defensible list of POs and transfer orders every week. Done poorly, it produces a Slack channel full of stockout screenshots.

This is not the same as buying for a seasonal drop. Drop buying is a one-shot bet. Replenishment is a repeating loop: forecast, monitor, reorder, receive, review. The loop only works if the data underneath it is trustworthy, which is where most apparel brands quietly break.

Why does replenishment planning matter for apparel brands specifically?

Apparel has a structural problem that masks replenishment failures: size and color. A style is not a SKU. A style is typically eight to twenty SKUs once you fan it out across the size run and colorway. You can have 90% sell-through on a style and still be out of stock on the medium black, which is the SKU that actually drove the revenue. Aggregate inventory looks healthy. The customer sees a sold-out button.

From the go-lives I have run this year, the pattern is consistent: brands between $5M and $100M discover that their replenishment logic is running at the style level when it needs to run at the SKU level, and that their forecast is built off shipped units rather than demand. Those two errors alone explain most of the chronic stockouts on hero programs.

Replenishment also matters because it sits directly on top of Breakpoint 3 in the 6 Breakpoints of Apparel Operations framework: inventory truth gets weaker. If you cannot trust inventory truthmber across DTC, wholesale, retail, and 3PL locations, no replenishment formula will save you. The math is correct. The inputs are lying.

How is demand-driven replenishment different from traditional reorder logic?

Traditional replenishment is static. You set a reorder point in January based on last year’s sales, and you revisit it when something breaks. The model assumes demand is roughly stable and lead times are roughly fixed. Neither is true in apparel. Wholesale at-once orders spike without warning. A single influencer post can drain a size run in 48 hours. Overseas lead times stretch by four weeks when a port congests.

Demand-driven replenishment uses live signals. It reads current sell-through velocity, channel mix, open wholesale orders, and committed transfers, then recomputes the reorder point and order quantity on a rolling basis. The shift from static to demand-driven is less a software upgrade and more an operating discipline. Someone owns the exception list. Someone explains why a SKU was reordered or skipped. The cadence is weekly, not quarterly.

My position: every apparel brand above $5M should be running demand-driven replenishment on at least its top 50 SKUs by revenue. Below that bar, the math of spreadsheets technically works. Above it, the spreadsheet becomes a liability the first time a planner takes a week off.

What are the core components of a replenishment planning system?

A workable replenishment system has five moving parts. Skip any one of them and the loop breaks within a season.

1. Demand forecasting

Forecasting estimates future units sold per SKU per channel per week. It uses historical sales, seasonality, promotional calendars, and known wholesale commitments. For apparel, the forecast must respect size curves. A style-level forecast that is then split by historical size ratio will under-order the meaty middle sizes every time.

2. Inventory level monitoring

You cannot replenish what you cannot see. Monitoring tracks on-hand, on-order, in-transit, allocated, and available-to-sell across every location: own warehouse, 3PL, retail stores, marketplace fulfillment. If those numbers live in different systems, your replenishment engine is guessing.

3. Reorder points

The reorder point (ROP) is the inventory level at which you trigger a new order. The formula is roughly: average daily demand multiplied by lead time in days, plus safety stock. The lead time number is where most brands lie to themselves. The contract says 45 days. The actual rolling average is 62. Use the actual.

4. Economic Order Quantity (EOQ)

Economic Order Quantity balances ordering cost against holding cost to find the order size that minimizes total cost per unit. For apparel, pure EOQ rarely survives contact with MOQs, fabric minimums, and container economics. Treat EOQ as a sanity check on the quantity your supplier or your container math is forcing on you, not as the answer itself.

5. Supplier management

Replenishment is only as reliable as the supplier behind it. Track on-time delivery, fill rate, quality rejection rate, and rolling lead time per factory. A supplier scorecard is not a procurement nicety. It is a replenishment input.

When does replenishment planning stop working?

Replenishment quietly fails at predictable thresholds. Three of them are worth naming.

The first is SKU count. Around 1,500 active SKUs, manual replenishment in spreadsheets stops being viable. You can build the model. You cannot maintain it. Updates lag by two weeks, and two weeks is a season in apparel.

The second is channel count. As soon as you add a third sales channel (say DTC, Shopify wholesale, and a marketplace), inventory allocation becomes a political question. Which channel gets the last 40 units of the medium? Without a replenishment plan that pre-commits stock by channel, the answer becomes whoever shouts loudest in Slack.

The third is geography. The moment you split inventory across two warehouses or add a 3PL, you have introduced transfers as a replenishment lever. Most brands forget that an inter-warehouse transfer is a replenishment decision and treat it as ad hoc logistics. It is not. It belongs in the same weekly review.

These thresholds map directly to Breakpoints 3, 4, and 5 in the framework: inventory truth, order flow trust, and warehouse execution predictability.

What techniques actually produce lean inventory?

Lean inventory is the outcome, not the technique. The techniques that produce it are unglamorous.

ABC segmentation. Sort SKUs into A (top 20% of revenue), B (next 30%), and C (the long tail). Run tight, demand-driven replenishment on A items weekly. Run looser, rule-based replenishment on B items biweekly. Make C items order-to-demand or run them out. Most planners spend equal attention on all three. That is the leak.

Size curve discipline. Hold a documented size curve per program. Re-fit it every season against actual sell-through, not against the curve your supplier prefers to cut. A 2-3-3-2 curve assumed against a real 1-3-4-2 sell-through guarantees lost sales on size large for the entire life of the program.

Lead time buckets. Group SKUs by actual lead time, not contractual. Domestic cut-and-sew at 21 days behaves differently from overseas at 90 days. Replenishment frequency and safety stock should be set per bucket.

Min-max with seasonality overlays. A flat min-max ignores that winter coats do not need replenishing in May. Overlay a seasonality index on top of min-max so the trigger moves with the calendar.

Exception management. Stop reviewing the whole catalog. Review the exception list: SKUs that crossed their reorder point, SKUs whose velocity changed more than 25% week over week, SKUs whose supplier is late. Everything else should run on autopilot.

How does technology change replenishment planning?

Technology does not replace the planner. It removes the parts of the job that the planner is bad at, which is mostly arithmetic at scale.

AI and machine learning for demand forecasting read larger datasets than a human can hold in working memory. They detect non-obvious patterns: that a SKU sells faster the week after a specific marketing email, that a wholesale account always under-orders in Q1 then panics in Q2. The output is a better starting forecast. The planner still owns the override.

ERP integration matters because replenishment touches everything: PLM data for new style introductions, inventory management for on-hand truth, production for open POs, order management for committed demand, warehouse for in-transit. If those modules do not share a single product and inventory record, your planner is reconciling instead of planning. This is the core argument for running PLM, PIM, production, inventory, order, and warehouse on one system rather than seven.

Automated replenishment writes the draft PO. The planner reviews and releases. This is the right division of labor. Full automation without a human in the loop produces excellent POs for the wrong styles.

Real-time analytics kill the Monday morning report. If the team is rebuilding the same pivot table every week, the data is in the wrong place. A live perpetual inventory view, plus a weeks-of-supply dashboard, plus a stockout-risk list, replaces most of the meeting.

Cloud-based inventory matters less for the scalability story and more for the fact that your 3PL, your overseas production manager, and your wholesale ops lead can all see the same number at the same time. That alone resolves about half of the recurring inventory arguments.

What does a weekly replenishment cadence look like in practice?

A workable cadence for a $20M apparel brand looks like this. Monday morning, the system produces three lists: SKUs below reorder point, SKUs whose velocity shifted materially, and POs with slipping receipt dates. The planner reviews the first list against open POs and writes draft purchase orders or transfers. The planner reviews the second list and adjusts forecasts where the shift is real (not noise). The planner reviews the third list and either pushes the supplier or rebalances stock from another location.

By Wednesday, POs are released. By Friday, exceptions are closed. The whole loop takes one planner roughly eight to ten hours a week for a catalog of around 3,000 SKUs. If it is taking four times that, the inputs are wrong, not the planner.

This cadence is the operational expression of maintaining weeks of supply as a steering metric rather than a reporting metric.

How do you measure whether replenishment planning is working?

Four numbers tell the truth.

In-stock rate on A items. Should hold above 95% on hero SKUs. Anything below 90% means the reorder point or the supplier is broken.

Weeks of supply by SKU class. A items should sit between 6 and 10 weeks. C items above 16 weeks are candidates for markdown or discontinuation.

Forecast accuracy. Mean absolute percentage error (MAPE) at the SKU-week level. Anything under 30% on A items is workable. Over 50% means the forecast is decoration.

Inventory turn. For most apparel brands, four to six turns a year is healthy. Below three, capital is trapped. Above eight, you are probably losing sales.

If you are not reporting these four weekly, replenishment is happening by feel.

Where does replenishment planning intersect with the rest of the operation?

Replenishment is not a standalone module. It pulls from production (open work orders, factory lead times), from order management (excess inventory signals, committed wholesale), from warehouse (in-transit, allocated), and from reporting (sell-through velocity). If those data streams live in disconnected tools, the replenishment planner becomes a human integration layer.

This is the structural argument behind the 6 Breakpoints framework. Breakpoint 1 (product data fragmenting) feeds bad SKU masters into the replenishment engine. Breakpoint 2 (production drift) corrupts your lead time inputs. Breakpoint 3 (inventory truth) destroys your on-hand. Breakpoint 4 (order flow) hides committed demand. Replenishment sits downstream of all four. Fixing the replenishment formula without fixing those upstream breakpoints is cosmetic.

What this means for an apparel operations team

For a $5M to $100M apparel brand, replenishment planning is the discipline most likely to be quietly outsourced to a senior planner’s intuition. That works until the planner takes a vacation, or the SKU count crosses 1,500, or you add a third sales channel. Then it does not.

The practical move is to stop treating replenishment as a spreadsheet routine and start treating it as a system: SKU-level forecasts, real lead times, weekly exception reviews, four steering metrics, and one source of truth for on-hand. Uphance exists to collapse those inputs onto one product, inventory, and order record so that the planner spends time deciding rather than reconciling. The goal is not perfect forecasts. The goal is clarity, not chaos, on the SKUs that actually drive the P&L.