Choosing the Right Replenishment Approach

ABC/XYZ segmentation

Segment items to align policy complexity with business value.

• ABC (value/velocity): A = high revenue/margin/critical, B = mid, C = low

• XYZ (variability): X = stable demand, Y = moderate variability, Z = intermittent/erratic

• Combine segments (e.g., AX, CZ) to drive differentiated policies and service targets

Demand and lead-time variability

• Coefficient of variation (CV) flags volatility; higher CV requires higher buffers or different policies

• Lead-time mean and variability matter as much as demand; unreliable lead times demand more safety stock or earlier ordering

• Seasonality and trend call for adaptive parameters and periodic review

Policy options: min–max, EOQ, periodic review, Kanban

• Min–max (continuous review): Replenish up to max when on-hand + on-order ≤ min (ROP)

• EOQ (fixed order quantity): Optimize order size for stable demand and cost structure

• Periodic review (order-up-to): Review every T; order to a target level covering T + lead time

• Kanban (pull): Visual/card-based triggers sized by demand and lead time; great for short lead times and stable usage

• Hybrids: Base-stock with EOQ caps, dynamic review periods, or service-level-driven min–max

Method selection matrix

Use a simple decision tree:

• Stable demand, reliable lead time, low order cost: EOQ or min–max

• Stable demand, batching/transport windows: Periodic review with order-up-to targets

• Variable demand, moderate lead-time variability: Min–max with service-level safety stock

• Intermittent demand (CZ/ZZ): Periodic review with demand-interval models or tailored reorder points (e.g., Croston/SBA for forecasting)

• Very short lead time, line-side replenishment: Kanban or two-bin

• Highly constrained (MOQs, capacity, supplier calendars): Periodic review with consolidation and order minimums

Constraints: MOQs, capacity, supplier calendars

• Respect MOQs, case packs, pallet layers, and truckload economics

• Align with supplier order windows, production cycles, and port cut-offs

• Consider slotting, labor, and dock capacity when setting review cadences

• Sustainability levers: Consolidate into fuller loads, reduce ship frequency, and position inventory to cut miles without harming service

Core Inputs and Formulas

Forecast and error metrics

• Forecast: Baseline statistical forecast enhanced by promotions, events, and sales input

• Error: Track MAPE/WMAPE, MAE, and bias at the planning level

• Variability: Use standard deviation and CV; for intermittent demand, model demand size and intervals separately

Lead-time modeling

• Measure average lead time and its standard deviation by SKU–supplier–lane

• Use in-transit and ASN/ETA data to update remaining lead time dynamically

• Distinguish systematic shifts (calendar, port delays) from random variation

Safety stock calculation

Safety stock buffers demand and lead-time uncertainty.

• If demand and lead time vary: SS = z × σLT

• σLT ≈ sqrt(σD^2 × L + (μD^2 × σL^2))

• Where z = z-score for target cycle service level, μD = mean demand per time unit, σD = demand standard deviation, L = mean lead time (time units), σL = lead-time standard deviation

For intermittent items, consider service on demand occurrences and simulate (or use Poisson/negative binomial) to set SS.

Reorder point formula

• ROP = expected demand during lead time + safety stock

• Demand during lead time ≈ μD × L

• In periodic review with review period T: Order-up-to S = μD × (L + T) + SS

EOQ and lot sizing

• EOQ = sqrt((2 × D × S) / H)

• D = annual demand units, S = order cost per order, H = annual holding cost per unit

• Adaptations: MOQ caps, rounding to pack sizes, periodic order quantity (POQ), or time-phased lot-for-lot when forecasts are dynamic

Service level vs fill rate

• Cycle service level: Probability of no stockout in a cycle

• Fill rate: Proportion of demand fulfilled immediately from stock

• For skewed or intermittent demand, fill rate is the more customer-centric KPI; service level is easier to control via z but does not guarantee units filled

Struggling to manage inventory manually? Try our free inventory tools to model demand and stock requirements instantly.

End-to-End Process Workflow

Data preparation and item segmentation

• Clean item master: lead times, pack sizes, units, shelf-life, cost, and locations

• Classify items into ABC/XYZ and life-cycle stages (NPI, active, end-of-life)

• Validate supplier calendars, MOQs, and transport constraints

Forecasting cadence

• Monthly S&OP for horizons and constraints; weekly refresh for short-term changes

• Demand sensing with POS/online signals for near-term adjustments

• Separate baseline, promotions, and causal lifts; maintain audit trails for overrides

Policy setting and parameter maintenance

• Assign policies by segment; set target service levels by item criticality

• Calculate safety stock, ROP, EOQ/target stock; round to packs and respect constraints

• Parameter governance: Version control, effective dates, and owner signoff

Execution and exception management

• Generate order proposals daily/weekly based on policy triggers

• Manage exceptions: stockout risk, expedite/defer suggestions, MOQ/rounding impacts

• Rebalance across the network when demand shifts or supply slips

Review and continuous improvement

• Post-event diagnostics: root causes for stockouts, backorders, and excess

• Parameter health checks: drift vs. variability changes

• Quarterly policy reassessment by segment; refresh lead-time models using actuals

Technology and Data Architecture

Roles of ERP, APS, WMS, OMS

• ERP: Item master, purchase orders, receipts, accounting

• APS/Planning: Forecasting, inventory policies, optimization, simulation

• WMS: On-hand accuracy, constraints (space/labor), FEFO/lot tracking

• OMS: Promise logic, ATP/available-to-allocate across channels

Integration with POS, supplier, and transportation data

• POS/eCommerce: Near-real-time demand signals and cannibalization detection

• Supplier portals/EDI: Forecasts, confirmations, ASNs, OTIF metrics

• Transportation/TMS: ETA, capacity, consolidation opportunities, lead-time history

• IoT/telemetry: Condition monitoring for perishables and in-transit visibility

AI/ML for demand and lead-time estimation

• Use ML for pattern recognition (seasonality, promotions), demand transference, and lead-time prediction

• Feature engineering: calendar events, weather, price, digital traffic, capacity signals

• Guardrails: Bias monitoring, cold-start strategies for NPIs, and explainability

Automation, alerts, and replenishment triggers

• Exception-based planning: Focus on high-impact SKUs, risk thresholds, and scenario alerts

• Workflow automation: Order release, supplier follow-up, and escalations

• Triggers: ROP/ROP breach, order-up-to reviews, and service-risk forecasts with confidence intervals

Multi-Echelon and Omnichannel Planning

DC-to-store allocation logic

• Fair-share allocation when supply is short; weight by demand, priority, and presentation minimums

• Respect pack sizes, display minimums, and shelf capacities

• Use allocation “guardrails” to protect top stores and avoid starving long-tail locations

Store ordering vs central replenishment

• Centralized auto-replenishment reduces noise and bias; stores provide event intel

• Store-ordering suitable for fresh and hyper-local items with short shelf-life

• Incorporate planograms and minimum display into target stock levels

E-commerce and drop-ship flows

• Virtual inventory pools across DCs and stores; ship-from-store with guardrails

• Promise against reliable ATP with safety buffers for lead-time risk

• Use drop-ship for long-tail to expand assortment without holding inventory

Pooling, transshipment, and risk positioning

• Inventory pooling reduces safety stock via demand aggregation

• Planned transshipments: Short-term rebalancing between nearby nodes

• Risk positioning/postponement: Keep upstream inventory generic; delay final configuration

Special Scenarios and Playbooks

New product introductions

• Use attribute/analog modeling and market tests to seed initial policies

• Start with conservative service targets and short review cycles; tighten as signals arrive

• Sunset plan: Controlled ramp-down with returns or markdown strategy

Promotions and events

• Uplift modeling with prebuild and forward deployment to key nodes

• Account for cannibalization and halo; coordinate allocation freezes before launch

• Post-event: Rapid sell-down plan, transfers, or returns to avoid excess

Perishables and shelf-life

• FEFO enforcement, spoilage modeling, and dynamic target stocks by day of week

• Balance service vs waste with shorter review periods and smaller lots

• Temperature and dwell-time monitoring; vendor lead-time adherence is critical

Long/variable lead times and imports

• Earlier order cut-offs, bigger lots, and calendar-based periodic review

• Buffer with in-transit inventory; use DC safety stock not store-level

• Container optimization: Cube fill, freight consolidation, and emissions impact

Supplier collaboration: VMI/CPFR

• VMI: Supplier maintains agreed SLAs and min–max; share POS/stock and rules

• CPFR: Joint forecasting, promotion planning, and exception reviews

• KPIs: OTIF, confirmation lead time, forecast consumption, and inventory turns

KPIs, Diagnostics, and Targets

Service level and fill rate

• Track service level by item-location cycle; set higher targets for A/critical SKUs

• Monitor unit fill rate to capture customer experience

• Use stockout cause codes to separate demand surges from supply failures

Inventory turns and days on hand

• Turns = annualized COGS / average inventory; DOH = 365 / turns

• Target ranges by segment (e.g., A/X higher turns than C/Z)

• Highlight slow movers and write-down risk early

Stockout and backorder metrics

• Stockout rate, lost sales, backorder duration, and partial-fill counts

• Split “avoidable” vs “unavoidable” stockouts for continuous improvement

• Tie expedites and penalties to root causes

Forecast accuracy and bias

• WMAPE by segment and horizon; track bias (positive/negative) separately

• Parameterize safety stock to actual forecast error distributions

• Use prediction intervals to inform z-level choices

OTIF and supplier reliability

• Measure on-time to request and to promise; in-full by line and unit

• Lead-time adherence and variability trends by lane

• Use scorecards to strengthen contracts and replenishment calendars

Implementation Roadmap and Change Management

Maturity model and quick wins

• Stage 1: Clean data, basic min–max for A/X; weekly review

• Stage 2: Segmented policies, EOQ and periodic review; exception dashboards

• Stage 3: Multi-echelon optimization, AI sensing, and automated ordering

• Quick wins: Fix bad lead times and pack sizes, prioritize top 20% SKUs, eliminate chronic MOQs misalignment

Pilot design and phased rollout

• Choose a focused category/region with clear constraints

• Baseline KPIs; run A/B against control for 8–12 weeks

• Exit criteria: Service up, inventory down, expedites reduced; then scale

Governance and ownership

• RACI for forecast, parameters, exceptions, and supplier engagement

• Monthly parameter council; quarterly policy review by segment

• Audit trails for overrides and event assumptions

Training and S&OP alignment

• Playbooks for NPI, promotions, and long-lead imports

• Cross-functional reviews: Merchandising, supply, logistics, finance

• Share scenario results and trade-offs in S&OP to set realistic targets

Worked Example and Tools

Sample calculation: safety stock, ROP, EOQ

Assumptions (weekly units):

• Mean demand (μD) = 500; demand standard deviation (σD) = 120

• Mean lead time (L) = 3 weeks; lead-time standard deviation (σL) = 1 week

• Target cycle service level = 95% → z ≈ 1.65

Calculations:

• Demand during lead time = μD × L = 500 × 3 = 1,500 units

• σLT = sqrt(σD^2 × L + μD^2 × σL^2)

• σLT = sqrt(120^2 × 3 + 500^2 × 1^2) = sqrt(43,200 + 250,000) ≈ 541 units

• Safety stock = z × σLT ≈ 1.65 × 541 ≈ 892 units

• Reorder point (ROP) = demand during L + SS ≈ 1,500 + 892 = 2,392 units

EOQ example (annualized):

• Annual demand D ≈ 500 × 52 = 26,000 units

• Order cost S = 75 per order; holding cost H = 2 per unit-year

• EOQ = sqrt((2 × 26,000 × 75) / 2) = sqrt(1,950,000) ≈ 1,396 units

• Round to pack size and respect MOQs and truck constraints

Cost–service trade-off illustration

With the above σLT:

• 90% service (z ≈ 1.28): SS ≈ 693; inventory lower, more backorders

• 95% service (z ≈ 1.65): SS ≈ 892; balanced outcome

• 98% service (z ≈ 2.05): SS ≈ 1,109; higher carrying cost, fewer stockouts

Tips:

• Simulate stockout costs and expedite fees to find the economic service target

• Use scenario analysis for supplier variability, promotion uplifts, and review frequency

Checklist and parameter review template

Monthly checklist:

• Refresh lead-time mean/variance by lane

• Recompute safety stock for items with >15% change in demand variability

• Validate ROPs vs. shelf capacity and presentation minimums

• Check MOQs/pack sizes and rounding losses

• Review top exceptions: stockout risk, chronic excess, expedite hotspots

• Align with upcoming promotions, events, and supplier shutdowns

Downloadable calculator and data schema

Build a simple spreadsheet calculator with columns:

• SKU, location, policy, review period (T), lead time (L, σL), demand mean (μD), demand stdev (σD), service target (z), SS, ROP, EOQ, MOQ, pack size
  Formulas:

• SS = z × sqrt(σD^2 × L + μD^2 × σL^2)

• ROP = μD × L + SS

• S (order-up-to) = μD × (L + T) + SS

• EOQ = sqrt((2 × D × S) / H)

FAQs

What is the difference between replenishment planning and demand planning?

• Demand planning produces the forecast and promotional assumptions.

• Replenishment planning turns those into inventory policies and orders, accounting for lead times, constraints, and service targets.

How do I set safety stock and reorder points?

• Choose a target service level by segment.

• Estimate demand and lead-time variability; compute SS = z × σLT.

• Set ROP = demand during lead time + SS, and round to pack sizes/MOQs.

Which is better: min–max, EOQ, Kanban, or periodic review?

• It depends on demand variability, lead-time reliability, and constraints.

• Stable demand: EOQ or min–max. Batched ordering: periodic review. Short-lead, stable usage: Kanban. Intermittent demand: periodic review with tailored buffers.

How do I plan replenishment for seasonal or intermittent demand?

• Use seasonal profiles and shorten review cycles near peaks.

• For intermittent items, model demand size and interval; consider Croston/SBA forecasting and periodic review with higher SS.

How often should replenishment parameters be reviewed?

• Recompute critical parameters monthly for A/volatile items; quarterly for stable C items.

• Recalculate immediately after major changes in lead time, demand mix, or supplier performance.

How should promotions and new product launches be handled?

• Build uplift forecasts with prebuild plans and allocation guardrails.

• For NPIs, use analogs and tight review cycles; adjust quickly as signals arrive.

What KPIs best measure replenishment performance?

• Service level, unit fill rate, inventory turns/DOH, stockout/backorder metrics, WMAPE and bias, OTIF, and supplier lead-time adherence.

What should I look for in replenishment planning software?

• Segmented policy support, multi-echelon capabilities, robust parameter governance

• Integration to POS, supplier, and TMS data; AI for demand/lead time

• Exception-based workflows, simulation/scenario planning, and auditability

Learn why growing Shopify brands choose Verve AI to automate forecasting, replenishment, and purchase planning.