Check the record first: inventory optimization for e-commerce
Inventory optimization turns stock data into a decision for each SKU and location. An operator checks demand, available quantity, supplier lead time, and buying constraints. The result may be a transfer, a replenishment, a pause in purchasing, or an investigation. No order is placed until someone approves it.
The essentials
- Inventory optimization starts with one SKU and one location. The operator verifies the stock record, then chooses whether to buy, transfer, pause, or investigate.
- A candle maker had 700 kg of usable wax that looked unavailable. A bill-of-materials rule let every scent draw on the shared wax.
- Do not buy when physical, reserved, and available quantities disagree. Trace the last sale, reservation, movement, and receipt before changing the order.
- A purchase order is one possible outcome. Repeated demand may trigger replenishment; a location imbalance may call for a transfer.
- myFulfillment shows stock, sales-history windows, supplier lead times, and suggested quantities. An operator can change or reject a suggestion before confirming a purchase.
On Monday morning, a candle-components manufacturer in southern France had one metric ton of wax in the building. Each scent needed 99% base wax and 1% fragrance oil. The team had assigned 300 kg to vanilla, 300 kg to exotic, 300 kg to berry, and 100 kg to a fourth scent.
Vanilla sold out on the storefront while 700 kg of the same wax remained on site. The warehouse count was right, but the selling rule hid usable stock. An operator can fix that inventory optimization problem.
What is inventory optimization?
Inventory management records each item, its location, and the units reserved for open orders. A stock-keeping unit, or SKU, identifies one sellable product. Inventory optimization helps the operator decide what to do with that SKU at that location.
An operator checks demand, supplier lead time, and buying constraints. Lead time runs from order to receipt. A named reviewer then approves a purchase, transfer, pause, or data investigation.
| Inventory management | Inventory optimization | |
|---|---|---|
| Question | What is on hand and available? | What should we do next? |
| Evidence | Movements, counts, reservations | Plus sales, lead time, constraints |
| Output | A corrected stock record | An action, owner, and date |
| Main risk | Acting on stale quantities | Treating a recommendation as an order |
Physical stock sits in the warehouse. Reserved stock is promised to an open order. Available quantity is physical stock minus reserved stock. If these quantities disagree, fix the record before following the e-commerce replenishment method.
Verify the stock record before you act
Before changing an order, trace the events behind the stock quantity. Nicole DeHoratius and Ananth Raman examined nearly 370,000 records across 37 physical stores in 2008. They found 65% were inaccurate (Management Science).
In a 2024 Institute for Supply Management article, CAPS Research reported average inventory accuracy of 91% and 67% for the lowest performers. ISM’s Jim Fleming proposed performance thresholds: "90 percent is good. And I would consider 95 percent world class" (ISM, The Monthly Metric: Inventory Accuracy Rate, 2024).
90%: good · 95%: world class
Accurate count, wrong sellable-stock rule
300 kg → 1,000 kg sellable quantity per scent
Channels out of sync, stale spreadsheet
Overselling stopped, same-day shipping
Yun Kang and Stanley Gershwin modeled stock that disappears without an inventory adjustment. The record can stay above the reorder point, the level that should trigger an order.
In their continuous-review model, one unit went missing for every 100 units demanded. Because the record never fell below the reorder point, the system placed no order, and 17% of demand went unfilled (Information Inaccuracy in Inventory Systems: Stock Loss and Stockout, IIE Transactions, 2005).
A dashboard may stay green until a customer reports the stockout. Read the last sale, order import, reservation, stock movement, and supplier receipt. The event history shows whether the quantity is safe to use.
When accurate stock still gives the wrong answer
The candle manufacturer had the correct total but the wrong sellable quantity. Sellable quantity is what the storefront can promise now. The original setup counted wax as finished-product stock before production.
| Field | What the record said | What was true |
|---|---|---|
| Storefront variant | Vanilla, out of stock | The same wax as every other scent |
| Base material | 300 kg allocated to vanilla | One metric ton of wax |
| Composition | A finished product | 99% base wax, 1% fragrance oil |
| Convertible within hours | No field in the old setup | 700 kg |
| Constraint | A stock shortage | A production step, not a purchase |
| Resulting action | Show the variant as out of stock | Publish the sellable quantity |
A forecast would not fix this setup. In a rollout still in progress, our team has connected the four storefronts to the warehouse and to the production line. The software now calculates sellable quantity from the bill of materials. A bill of materials (BOM) lists a product’s components.
Once the operator applies the BOM, the rule lets every scent draw from the full ton. Each variant’s sellable quantity moves from 300 kg to 1,000 kg (how the calculation works). Apply the same check to kits and supplier stock.
Which inventory optimization metrics should you review?
At the end of June 2026, U.S. retail trade held $832.6 billion in inventory. Its seasonally adjusted inventory-to-sales ratio was 1.25, down from 1.30 a year earlier. The U.S. Census Bureau reported both figures in Manufacturing and Trade Inventories and Sales, release CB26-132.
| Metric | What to inspect | Decision this evidence can change |
|---|---|---|
| Available quantity | Physical minus reserved | Buy, transfer, or neither |
| Comparable sales window | Demand vs. seasonal reference | Is the spike real? |
| Days of supply | Stock vs. typical demand | Which SKUs come first? |
| Stockout frequency | Count and timing | Should the team act earlier? |
| Open purchase orders | Quantity, date, backorders | Should the buyer order? |
| Supplier lead time | Promised vs. actual receipt | When to review again? |
| Age and last sale | Receipt and sale dates | Pause or discontinue? |
| Availability by location | Available and reserved quantities | Transfer before buying? |
These measures can be misread. Physical stock may include reserved units. A promotion may look like normal demand, or a quoted lead time may hide actual delays.
ABC analysis ranks items for review. A forecast estimates demand from past sales, while safety stock covers uncertainty. SKU rationalization removes slow-selling products from the assortment. Just-in-time purchasing times orders to arrive when the stock is needed.
Choose a technique after two checks. The stock records must reconcile. The demand pattern must repeat.
Before buying, check available stock and open orders to confirm the shortage.
After both checks pass, confirm a shortage before buying. An ABC class does not settle the decision. A low-volume item may need close control when lead time is long or it requires a high service level.
| Symptom | Check first | Action and owner |
|---|---|---|
| Repeated stockout | Sales, lead time, open orders | Replenish sooner (buyer) |
| Aged stock | Age, last sale, season | Pause buying (buyer) |
| Uneven demand | Windows, campaigns, returns | Document the exception (buyer) |
| Long or variable lead time | Promised vs. actual receipts | Cover uncertainty (purchasing) |
| Location imbalance | Available and reserved quantities by site | Transfer first (operations) |
| Records do not reconcile | Quantities and timestamps | Count stock; buy nothing yet (operations) |
A supplier's minimum order quantity, or MOQ, is the smallest quantity it accepts. Pack size is the number of units ordered together. Either constraint can push a purchase above the SKU's need.
Create one spreadsheet row for each SKU and location. Add these evidence fields: SKU, location/channel, review date, available quantity, open orders/backorders, recent sales window, comparable seasonal window, supplier lead time, MOQ/pack, warning level, ideal level, known exception, and evidence confidence. Add these decision fields: proposed action, quantity, reviewer, decision date, and next review.
A false shortage can prompt an unnecessary order. Record any MOQ or pack-size exception beside the quantity. The weekly list should explain each alert; "available quantity below warning level" is clearer than a red cell.
Add a timestamp to the evidence. Monday's balance should not support a Friday order without another check.
Two teams changed where they checked availability, not how much they held
A multi-brand lifestyle retailer runs 12,000+ SKUs across one web store, one large marketplace, and two niche marketplaces. Its buyer spent two full days each week using a spreadsheet to review the last 30 days of sales. At peak, top sellers went out of stock 15 to 20 times each month, while winter products accumulated.
We rebuilt the review around sales velocity and the same period last year. The two-day task became a 30-minute review. Stockouts on top sellers fell to zero within the first quarter (customer results).
The buyer brought chronic overstock on slow movers under control. The buyer said: "No more emergency supplier calls. The system tells us what to order before we even think about it."
The buyer still approves every order. The software saves time by assembling evidence, not by deciding (how the replenishment was rebuilt).
A French home-and-garden retailer could not maintain one stock figure across its channels. It sold through its web store, three marketplaces, and a B2B portal. Staff updated the retailer's spreadsheet when they had time.
When a product sold on one channel, the stock shown on the others did not fall. The retailer oversold on a marketplace and shipped one to two days late. We centralized the multi-channel order flow.
Overselling stopped, and the team went from five platform tabs to one screen. Same-day shipping became standard. The team switched to a single source for availability. Its physical stock did not change.
From an inventory decision to a purchase order
Procurement begins once the operator chooses to replenish. The buyer then decides when to order and how much to buy from the supplier.
For a confirmed shortage, the replenishment guide explains sales velocity, reorder point, safety stock, and Min-Max rules. The buyer checks open orders, MOQ, and pack size. A named approver can then create or amend the purchase order.
In myFulfillment, Stock Helper lists three sales-history windows. The same screen shows available and physical quantities, quantities to ship, warning and ideal levels, and supplier lead times. Stock Helper calculates a recommended quantity and displays it to the operator.
The operator can change or reject the suggestion. The operator must confirm the supplier summary before the software creates a purchase order. The software does not place an order by itself.
- Evidence
Sales, availability, lead time and constraints
- Recommendation
Proposed levels and quantity
- Operator review
Check, adjust or reject
- Confirmed action
Buy, transfer, pause or investigate
- Monitored result
Availability, cycle time or aged stock
myFulfillment assembles the evidence. A named operator decides and sets the next review date. Central stock visibility supports that decision. After approval, the buyer can use the supplier workflow to create a purchase order.
Start with the SKU that caused the most disruption this month. Reconcile physical, reserved, and available quantities. Then check open orders against promised lead time and record the action, owner, decision date, and next review.
Sources
- Nicole DeHoratius and Ananth Raman, Inventory Record Inaccuracy: An Empirical Analysis, Management Science, 2008 (DOI 10.1287/mnsc.1070.0789).
- Yun Kang and Stanley B. Gershwin, Information Inaccuracy in Inventory Systems: Stock Loss and Stockout, IIE Transactions, 2005 (DOI 10.1080/07408170590969861; linked text is the MIT preprint).
- Institute for Supply Management, The Monthly Metric: Inventory Accuracy Rate, 2024, reporting The Metrics of Inventory Management (2023) by CAPS Research (Tempe, Arizona).
- U.S. Census Bureau, Manufacturing and Trade Inventories and Sales, release CB26-132, seasonally adjusted preliminary June 2026 estimate, released August 2026.
- Boostmyshop client cases, anonymized, 2026: artisan candle manufacturer, multi-brand lifestyle retailer, multi-channel retailer.
Frequently asked questions
An operator uses inventory optimization to choose the next action for one SKU at one location. The operator checks demand, available quantity, supplier lead time, buying constraints, and service commitments. A reviewer then approves a replenishment, transfer, pause in purchasing, or data investigation.
Start with one SKU, one location, and one review date. Check available quantity, open orders, backorders, sales windows, supplier lead time, MOQ, pack size, and exceptions. Record the evidence, action, reviewer, and next review date.
Match the technique to the problem. Use ABC analysis to prioritize reviews, forecasting to test demand, safety stock for uncertainty, SKU rationalization for slow sellers, and allocation for location imbalances. Repair the stock record before applying any method.
Set the schedule based on the SKU’s risk profile. Review items with variable demand, seasonal sales, or long lead times more often. Include items with a high service commitment. Review again after a supplier delay, promotion, listing change, stock-count correction, or repeated stockout. Record the next date on each decision.
Choose a tool that shows its inputs for each SKU and location. It should expose open orders, supplier constraints, and exceptions. The operator must be able to understand the recommendation and approve it before anyone places an order.
Multi-echelon inventory optimization coordinates stock across linked stages or locations. A smaller e-commerce team can apply the principle by checking channel and warehouse availability before buying. Complex network models sit outside this one-SKU method.
Test a small set first, including one exception. Compare each sales window with its source orders. Confirm whether lead time belongs to the SKU or comes from a supplier default. Apply levels in bulk only after those checks pass.
Continue exploring
Stock availability, picking accuracy, and shipping times affect marketplace sales. Jonathan Coloma, COO of Repiauto, explains how his team went from 300 to 1,000 orders a day in two years. He describes the limits of paper-based work, ERPs built for different needs, a go-live estimate of fifteen days, and changes to his daily routine.
To hold up under peak volume, split the order queue into waves based on real operating constraints, then decide where barcode checks belong. At the packing bench, the operator verifies the parcel. Carrier rules select the service and print the label, so the next order can start without another decision.
Industrializing your purchasing means ordering on rules, not guesswork: at the right time (the reorder point), in the right quantity (your coverage period), from the right supplier, with a purchase order validated in one click. Available on replay.
A higher conversion rate becomes profitable growth only when your logistics operation can handle the additional order volume. In 45 minutes, identify the pressure points that put inventory accuracy, order picking, shipping, customer support and margin at risk as sales rise. The webinar is presented in French.
Let’s review your order workflow
In 20 minutes, we’ll identify bottlenecks and ways to save time. No slides.
Free 20-min call · No commitment