Inventory Management for Online Stores: A Practical Guide

The foundation: SKU discipline

Before any system or report, every sellable item needs a unique, stable code. Unique means every variant (size, colour, capacity) has its own code, not the parent product's. Stable means the code never changes once used — changing it severs the item's history and breaks any existing integration.

Keep codes human-readable and consistent: Latin letters and digits, no spaces or special characters, following one pattern such as category-product-colour-size. It looks like housekeeping, but everything else rests on it — including connecting your store to a fulfillment provider.

Stock accuracy: one source of truth

The most common failure in growing stores is having several balances: one in the store, one in a spreadsheet, and one in the warehouse manager's head. The fix is not making each of them more accurate — it is choosing one source of truth from which the others derive.

In practice: the warehouse system is authoritative, the store mirrors it automatically, and any manual correction goes through a recorded movement (receipt, issue, adjustment) rather than someone overwriting a number. A recorded movement leaves a trail you can audit; an overwritten number leaves nothing.

Counting: cyclical, not annual

A full annual count halts operations and surfaces errors long after they mattered. Cycle counting is better: count a small rotating subset of SKUs each week so everything is covered within a defined cycle, without stopping work.

Prioritise by importance rather than alphabetically: fast-moving, high-value items get counted more often, slow items less. Measure the outcome with one clear metric — the percentage of locations where physical stock matched the system — and watch its trend rather than a single snapshot.

When to reorder: reorder point and safety stock

Stockouts are not sudden; they happen because nobody calculated when to order. The base rule:

Reorder point = (average daily demand × lead time in days) + safety stock

Lead time is not what the supplier promises — it is what actually happens, from raising the purchase order until the item is sellable in your store, including production, freight, customs clearance, and warehouse receiving. Ignoring those stages is the number one cause of surprise stockouts.

Safety stock protects against demand variability and supply delay. Do not apply one flat number across all SKUs: volatile items or those with unreliable suppliers need a wider buffer; stable, quickly replenished items need less.

Do not treat all SKUs alike: ABC analysis

In most stores a small subset of items generates most of the revenue. Rank your SKUs by revenue contribution and split them into three classes:

• A: the items driving most of your revenue — close monitoring, frequent counts, never allowed to run out.
• B: moderate contribution — periodic review and a moderate buffer.
• C: small contribution despite large SKU count — order in larger, less frequent batches, and seriously question whether some should stay in the catalogue.

The value of the classification is that it directs limited attention where it changes outcomes, instead of spreading it evenly across everything.

Dead stock: the silent loss

An item that has not moved in a long time costs you three ways: frozen capital, paid storage space, and value that decays over time. Define a clear ageing threshold suited to your product (say, no sales in 90 or 180 days), review the list monthly, and make an explicit call — discount, bundle, or liquidate.

Deferring that decision is itself a decision to lose money, just an unannounced one.

Warehouse organisation affects your balance

Stock accuracy is not purely a software matter. An item scattered across unrecorded locations, or stored next to something that looks similar, produces picking errors that show up later as count variances. Every location has an identifier, every item is bound to its location, and every movement is recorded — that is what makes a balance trustworthy.

For items with expiry dates or batches, apply first-in-first-out so the oldest batch is picked first, and capture the batch number at receiving.

Five metrics are enough to start

• Stock accuracy: share of locations matching between system and reality.
• Inventory turnover: how many times stock sold through and replenished in the period.
• Days of cover: how many days your current balance lasts at the current sales rate.
• Stockout rate: how often an item was unavailable when demanded.
• Dead stock share: non-moving items as a proportion of total inventory value.

Five numbers reviewed monthly beat twenty reports nobody opens.

Where to start if nothing is organised

Start with SKU coding, then one full count to establish a trustworthy baseline, then a single source of truth that is never manually overwritten, then cycle counting to hold accuracy. Do not attempt all of it in a week — the order is not a suggestion, since each step depends on the one before it.