Inventory Management for SMEs: Service, Stock and Cash

A practical inventory course for owners, buyers, planners and operations managers in small and mid-sized distributors and manufacturers. You will learn what inventory really costs, how to measure it, how to classify items with ABC and XYZ analysis, how to forecast simply and measure forecast error, and how to set reorder points, safety stock and order quantities. It closes with supplier collaboration, MRP basics for manufacturers, system and data discipline, and a 90-day improvement plan.

Course structure

• Module 1: Inventory Basics (4 lessons). Why inventory matters, what it costs, the types of inventory and their purposes, the measures that matter and how to keep records accurate.
• Module 2: Classifying Items (4 lessons). ABC analysis by value, XYZ analysis by demand variability, policies for each class and how to handle slow-moving and obsolete stock.
• Module 3: Demand and Forecasting (4 lessons). Demand patterns, simple forecasting methods, forecast error and bias, and a monthly sales and operations planning routine.
• Module 4: When and How Much to Order (4 lessons). Reorder points and real lead times, safety stock, economic order quantity and the main replenishment methods.
• Module 5: Improve and Sustain (4 lessons). Supplier collaboration, MRP and bill-of-materials accuracy, systems and data discipline, and a 90-day improvement plan.

Interactive labs

• Why is this stock here?: Sort twelve situations from a distributor and a manufacturer by the purpose of the stock, then check your answers. Knowing why stock exists tells you which lever reduces it.
• Value and variability: where each item belongs: Place seven items on a grid of annual value against demand variability to see which need tight control, which suit simple rules and which need customer collaboration.
• Safety stock, reorder point and order quantity: Enter demand, variability, lead time, service level and costs for one item to calculate safety stock, reorder point, economic order quantity and average inventory value.

How it works

• Work through 20 short lessons at your own pace. Each has key takeaways, an exercise for your own workplace and a quick knowledge check.
• Try the three interactive labs as you reach them in the course.
• Pass the 20-question final exam (70% or higher, unlimited retakes) to unlock your Chmltech certificate of completion.
• Free, with no sign-up. Your progress is saved in this browser.

Inventory sits between what you buy and what you sell. Too little and you miss orders, expedite freight and stop production lines. Too much and cash is tied up on shelves, space fills, and stock ages, gets damaged or becomes obsolete. Good inventory management is the balance between service (having what customers need when they need it) and cost (the money and effort tied up in stock).

What inventory really costs

The price paid for an item is only the start. Holding it costs money every day it sits there. Carrying cost includes:

Annual carrying cost is often estimated at around 20 to 30 percent of average inventory value, and higher for perishable, fashion or fast-changing technical products. Estimate your own: even a rough figure changes decisions. At 25 percent, an extra $200,000 of stock costs about $50,000 a year to hold.

The cost of running out

Stock-outs cost too: lost sales, expedited freight, overtime, idle production, partial shipments and, most of all, customer trust. Unlike carrying cost these rarely show up as one line in the accounts, which is why many businesses overstock "to be safe". This course gives you methods to decide how much is enough, item by item.

Who owns inventory?

Inventory touches sales (forecasts and promises), purchasing (what and when to buy), operations (what to make), the warehouse (accuracy) and finance (cash). Problems often come from these groups working to different goals. A shared set of measures and a regular planning meeting, covered later in the course, align them.

Inventory is classified two ways: by form and by purpose. Both matter, because each purpose calls for a different decision.

When someone says "we have too much inventory", ask which kind. A warehouse full of anticipation stock before a seasonal peak is planned and healthy. The same space full of slow movers from discontinued lines is a problem. Splitting inventory by purpose turns a vague complaint into specific actions.

The lab after this lesson asks you to classify examples by purpose.

You cannot manage inventory by feel. A short set of measures, tracked monthly and by item class, shows whether you are improving.

Example

A distributor's annual cost of goods sold is $6 million and its average inventory is $1.5 million at cost. Turns are 6,000,000 ÷ 1,500,000 = 4, so it holds about 365 ÷ 4 ≈ 91 days of stock. If better ordering lifts turns to 5 at the same sales, average inventory falls to $1.2 million, freeing $300,000 of cash and saving the carrying cost on it.

Use measures in pairs

Turns and fill rate pull in opposite directions: cutting stock raises turns but can lower fill rate. Always report them together, and by item class, so no one improves one by quietly damaging the other. Compare yourself with your own history first; industry averages vary too widely to be a reliable target.

Every inventory method in this course depends on the system's quantities being right. If the system says 40 and the shelf has 12, the reorder point never triggers and you run out. If the shelf has 80, you buy more of what you already have. Inventory record accuracy is the foundation.

Why records drift

• Receipts not entered, or entered against the wrong item or location.
• Picks taken from a different location than the system says.
• Unit-of-measure confusion: cases versus eaches, metres versus rolls.
• Scrap, samples and returns not recorded.
• Production consumption that differs from the bill of materials.

Cycle counting instead of one big count

An annual wall-to-wall count shuts operations for a day or more, finds errors months after they happened and fixes the numbers but not the causes. Cycle counting counts a small set of items every day or week, so every item is counted several times a year, more often for important items. Typical frequencies: A items monthly or quarterly, B items quarterly or twice a year, C items once or twice a year. Many firms also count when a location reaches zero, since it is quick and catches errors early.

Make the counts improve the process

The real value of cycle counting is finding why errors happen. For each discrepancy above a tolerance, investigate the recent transactions, find the cause and fix it: a missing receiving step, an unclear label, a picker using the wrong unit. Measure accuracy as the percentage of counts within tolerance. Many organizations aim for 95 percent or better on all items and higher on A items.

Make counts blind (the counter does not see the system quantity), and separate counting from the person who adjusts the record.

Not all items deserve the same attention. ABC analysis ranks items by annual usage value (annual units used or sold × unit cost) and groups them:

The exact cut-offs are a choice, not a law. Pick ones that give you a manageable number of A items.

How to do it

1. Export 12 months of usage or sales by item, with unit cost.
2. Calculate annual usage value for each item.
3. Sort from highest to lowest.
4. Add a running total and a cumulative percentage of total value.
5. Assign classes at your chosen cut-offs, for example A up to 80 percent cumulative value, B up to 95 percent, C the rest.

Then adjust with judgement. A low-value item that stops a production line or is critical to a key customer may need A-level care. Many firms add a criticality flag rather than changing the ABC class itself.

Repeat the analysis every quarter or twice a year: products move between classes as sales change. The Chmltech Insights series walks through full worked examples for a distributor and a manufacturer, with a downloadable Excel template.

ABC tells you how much an item is worth. It does not tell you how hard it is to plan. XYZ analysis classifies items by how variable their demand is, usually using the coefficient of variation (CV): the standard deviation of demand per period divided by the average demand per period.

As with ABC, the cut-offs are a choice. Use the same period length for every item, usually months, and at least 12 periods of history.

Combining ABC and XYZ

The nine-cell ABC-XYZ grid gives each group a sensible policy:

AZ items, high value and erratic demand, are the hardest. Holding enough safety stock for erratic demand on expensive items costs a lot. Better options are working with customers on advance notice, agreeing lead times, or not stocking at all. The lab after this lesson lets you place items on a value and variability grid.

Classes are only useful if they lead to different treatment. Write a one-page inventory policy that sets, for each class, the rules people should follow.

Why C items often get high service levels

It seems backwards, but it is usually right. C items are cheap, so extra safety stock costs little. Yet a missing C item, such as a fastener or a label, can still stop a shipment. Holding a bit more of C items and buying them less often frees the planners' time for A items, where every unit of stock is expensive.

Service level targets

Define service level clearly. Cycle service level is the probability of not running out during a replenishment cycle. Fill rate is the share of demand met from stock. They are not the same: a 95 percent cycle service level usually gives a fill rate well above 95 percent, because when you do run out it is often only for part of the demand. Use cycle service level to set safety stock and fill rate to report service to customers.

Every business accumulates stock it does not need: discontinued products, over-buys, customer-specific items left after a contract ended, engineering changes that made parts redundant. This excess and obsolete (E&O) stock costs space, handling and carrying cost, and the longer it sits the less it is worth.

Find it

Run a monthly report of items with no usage in the last 6 or 12 months, and items with more than a set number of months of supply, such as 12 months based on the last year's usage. Show the value and age of each.

Decide what to do

Involve finance early. Write-downs and write-offs affect the financial statements, and the timing and treatment may have tax implications that your accountant should advise on. Holding obsolete stock in the hope that its value returns usually costs more than acting.

Prevent it

• Review new-product and end-of-life plans with purchasing so buying stops in time.
• Check engineering changes for stock on hand and on order before they take effect.
• Agree customer commitments for customer-specific stock in writing.
• Avoid large buys for small discounts without checking months of supply.

Before choosing how to forecast or how much safety stock to hold, look at the shape of each item's demand. Plot 24 months of history by item. Most items show one or more of these patterns:

Clean the history first

Forecasts built on raw sales can mislead. Adjust for:

• Stock-outs: sales were zero because you had none, not because no one wanted it. Use orders or estimated lost demand where you can.
• One-off events: a single project order or a customer's one-time stock-up.
• Promotions: a price promotion that pulled demand forward.
• Calendar effects: months with different numbers of working days.

Dependent and independent demand

Finished products sold to customers have independent demand: you forecast it. Components used to make those products have dependent demand: you calculate it from the production plan and the bill of materials. Forecasting components separately from their parent products adds avoidable error. Manufacturers use material requirements planning (MRP) for dependent items, covered later in the course.

Small and mid-sized businesses rarely need complex forecasting software to start. A few simple methods, applied consistently, beat gut feel.

Exponential smoothing in one example

Last month's forecast was 200 units and actual demand was 240. With α = 0.2, the new forecast is 200 + 0.2 × (240 − 200) = 208. A higher α reacts faster to change but also to noise; a lower α is steadier. Values around 0.1 to 0.3 are common starting points.

Seasonal indices

If November demand has averaged 1.4 times the monthly average over several years, its seasonal index is 1.4. With an underlying level of 500 units a month, the November forecast is 500 × 1.4 = 700. Use at least two, ideally three, years of history to estimate seasonal indices.

Forecast at the right level

Forecasts are more accurate for groups than for single items, and for longer periods than shorter ones. Many firms forecast product families by month, then split to items by recent mix. Always keep a record of who overrode a forecast and why, so you can learn whether the overrides helped.

Every forecast is wrong; the question is by how much and in which direction. Measuring error tells you how much safety stock you need and whether your method is improving.

Example

Over four months an item had actuals of 100, 120, 90 and 110 against forecasts of 110, 110, 110 and 110. Errors are −10, +10, −20 and 0. MAD = (10 + 10 + 20 + 0) ÷ 4 = 10 units. Bias = (−10 + 10 − 20 + 0) ÷ 4 = −5 units a month: on average the forecast was 5 units too high.

Why bias matters most

Random error averages out; bias does not. A forecast that is always 15 percent high builds excess stock month after month. A forecast that is always low causes repeated stock-outs and expediting. Bias often comes from people: sales teams rounding up to protect supply, or planners rounding down under cash pressure. Track bias by item group and by who made the forecast.

MAPE can mislead for low-volume items, because a miss of 2 units on actual demand of 1 is a 200 percent error. For intermittent items, use MAD or a weighted measure instead. Review error and bias monthly in the planning meeting covered in the next lesson.

Sales and operations planning (S&OP) is a monthly routine that brings sales, operations, purchasing and finance together to agree one plan for the next several months. Without it, sales forecasts one number, operations plans another, purchasing buys to a third and finance discovers the result in the bank balance.

A simple monthly cycle

In a small company this can be two meetings a month of an hour each, using a shared spreadsheet. The discipline matters more than the tools.

What the plan covers

• Demand by product family for the next 3 to 18 months.
• Production or purchasing plan and capacity, including overtime or extra shifts.
• Planned inventory by family, in units and dollars, compared with targets.
• Key risks: a large customer's uncertain order, a supplier's capacity, a planned shutdown.

The benefit is fewer surprises: purchasing sees peaks early, operations plans capacity, and finance sees the cash needed for inventory builds. The Chmltech article on S&OP for small manufacturers goes further.

The two core questions of replenishment are when to order and how much. The reorder point (ROP) answers the first: when stock on hand plus stock on order falls to the ROP, place an order.

ROP = demand during lead time + safety stock

If an item sells 40 units a working day and the supplier takes 10 working days to deliver, demand during lead time is 400 units. Without safety stock, you would order when 400 are left and expect to run out just as the delivery arrives. Safety stock, covered next, protects against demand being higher or the delivery being later than expected.

Lead time is more than transit

The lead time that matters is the full replenishment time, from the moment stock reaches the reorder point to the moment new stock is available to pick or use:

• Time to notice the need and raise the purchase order (review delay).
• Supplier processing and production time.
• Transit time, including customs for imports.
• Receiving, inspection and put-away.

Many companies use the supplier's quoted lead time and are surprised by stock-outs. Measure actual lead times from your purchase order and receipt dates, and track their variation as well as their average.

Use inventory position, not just on-hand

Compare the ROP with inventory position: on hand + on order − backorders and committed stock. Using only on-hand stock causes duplicate orders: the item stays below the ROP while the first order is in transit, so the system keeps suggesting more.

Safety stock is extra stock held to cover uncertainty: demand higher than forecast, or a delivery that arrives late. The question is how much, and the answer depends on how variable demand and lead time are and on the service level you want.

The standard formula (demand variability only)

Safety stock = z × σ_d × √L

• z is the service factor for the target cycle service level: about 1.28 for 90%, 1.65 for 95%, 1.96 for 97.5% and 2.33 for 99%.
• σ_d is the standard deviation of demand per period (for example per day), ideally of forecast error rather than raw demand.
• L is the lead time in the same periods.

Example: daily demand varies with a standard deviation of 12 units, lead time is 10 days and the target is 95%. Safety stock = 1.65 × 12 × √10 ≈ 1.65 × 12 × 3.16 ≈ 63 units. With average demand of 40 units a day, the reorder point is 400 + 63 = 463 units.

When lead time varies too

If lead times vary, safety stock must cover that as well. A common formula combines both sources: z × √(L × σ_d² + d² × σ_L²), where d is average demand per period and σ_L is the standard deviation of lead time. Unreliable suppliers can require much more safety stock than variable demand.

Why service level choices matter

Safety stock rises steeply as the target nears 100%. Going from 95% to 99% needs about 40% more safety stock (2.33 ÷ 1.65). Set high targets for items and customers where stock-outs are costly, and lower targets where they are not.

Review safety stocks at least quarterly. They are not set-and-forget numbers. For more worked examples see the Insights articles on setting safety stock and safety stock for manufacturers.

The reorder point says when to order. The order quantity says how much. Ordering large amounts rarely means fewer orders and less ordering effort but more cycle stock; ordering small amounts often means the opposite. The economic order quantity (EOQ) finds the balance.

EOQ = √(2 × D × S ÷ H)

• D is annual demand in units.
• S is the cost to place and receive one order: purchasing time, receiving, inspection and invoice processing. For manufactured items, the setup cost.
• H is the annual cost of holding one unit: unit cost × carrying cost rate.

Example: annual demand is 10,000 units, each order costs $60 to process and a unit costs $18 with a 25% carrying rate, so H = $4.50. EOQ = √(2 × 10,000 × 60 ÷ 4.50) ≈ √266,667 ≈ 516 units, about 19 orders a year.

EOQ is a starting point

The total cost curve is flat near the EOQ, so rounding to a convenient quantity costs little. Then adjust for real constraints:

• Minimum order quantities and pack sizes from the supplier.
• Price breaks: compare the saving on price with the extra carrying cost of the larger order.
• Shelf life and storage space.
• Freight: full pallets or full truckloads may be worth more than a small move away from EOQ.

Reduce the cost of ordering

The most powerful lever is not the formula but S. If blanket orders, supplier-managed replenishment or automated purchase orders cut the cost per order, EOQ falls, orders become smaller and more frequent, and cycle stock drops. The same logic drives quick changeovers in manufacturing. The lab after this lesson calculates safety stock, reorder point and EOQ together.

How often you check stock shapes how you order. There are three common approaches.

The trade-off of periodic review

With periodic review, safety stock must cover the review period plus the lead time, because you might drop just after a review and not order until the next one. Weekly review with a 10-day lead time means protecting about 15 working days, not 10. That means more safety stock, but it allows one consolidated order per supplier per week, which can cut freight and ordering effort.

Setting the target levels

• For periodic review: order-up-to level = demand over (review period + lead time) + safety stock for that period.
• For min-max: min is the reorder point; max is min plus a sensible order quantity.

Whichever method you use, the parameters must be recalculated as demand and lead times change. Parameters set years ago and never reviewed are one of the most common causes of both excess stock and shortages.

Safety stock and cycle stock both depend on suppliers: how long they take, how reliably they deliver and how small an order they accept. Working with key suppliers is often the biggest lever on inventory.

Measure supplier performance

Share the results with suppliers each quarter. Many delivery problems come from unclear expectations on both sides.

Levers to discuss

• Share forecasts for the next three to six months so suppliers can plan capacity.
• Blanket orders with scheduled releases cut ordering effort and secure capacity.
• Smaller, more frequent deliveries for high-volume items, sometimes on a fixed weekly run.
• Vendor-managed inventory (VMI) or consignment, where the supplier manages or owns stock at your site.
• Local or second sources for items with long, variable import lead times.

Total cost, not unit price

A cheaper overseas supplier with a 12-week, variable lead time may cost more in total than a local one with a 2-week lead time, once safety stock, cycle stock, freight, duties, expediting and stock-out risk are counted. Compare suppliers on total landed cost and inventory impact, not purchase price alone.

Manufacturers plan components differently from finished goods. Material requirements planning (MRP) works out what to buy and make, and when, from three inputs:

1. The master production schedule (MPS): which finished products to make, how many and when.
2. The bill of materials (BOM): which components, and how many of each, go into each product.
3. Inventory records: what is on hand and on order.

MRP explodes the MPS through the BOM to get gross requirements for each component, subtracts on-hand and scheduled receipts to get net requirements, and offsets by lead time to suggest when to release orders.

Example

A plant plans 500 cabinets in week 6. Each cabinet uses 4 hinges. Gross requirement: 2,000 hinges in week 6. With 600 on hand and 400 on order arriving in week 3, net requirement is 1,000. With a 3-week lead time, the purchase order for 1,000 hinges, rounded up to the supplier's pack size, must be released in week 3.

Garbage in, garbage out

MRP is only as good as its data. The most common problems:

• BOM errors: wrong quantities, missing parts or outdated revisions. Aim for very high BOM accuracy, verified by audits against actual builds.
• Inventory inaccuracy, covered earlier.
• Unrealistic lead times or schedules that exceed capacity.
• Unrecorded scrap and substitutions on the shop floor.

Planners who do not trust the system build spreadsheets and buffers around it, which raises inventory and hides the data problems. Fix the data, then trust the system. The Insights article on MRP covers the logic in more detail.

Inventory methods need a system to run them. For many SMEs that starts in accounting software or a spreadsheet and grows into an ERP system, sometimes with a warehouse management system (WMS) for larger or more complex warehouses.

Data discipline beats features

Most inventory systems underperform because of data, not software. Common gaps: missing or wrong lead times, replenishment parameters never updated, units of measure mixed up, and transactions recorded hours or days after the event. Assign an owner to the item master, set rules for creating and changing items, and review parameters on a schedule.

Before you buy a new system

Fix what you can in the current one first: clean the item master, add bin locations, start cycle counting and set parameters by class. You will learn what you really need, and a new system will start with clean data instead of migrating the old problems. When you do select a system, script demonstrations around your own processes and data rather than the vendor's standard demo.

Inventory improvement works best as a short, focused programme with clear measures. Here is a 90-day plan that fits most SMEs.

Set targets in pairs

For example: raise turns from 4 to 5 and keep fill rate at or above 97 percent; raise inventory accuracy to 95 percent; cut excess and obsolete stock by a third. Report progress monthly by class.

Common traps

• Across-the-board cuts: "reduce inventory 20 percent" without class rules usually cuts the wrong items and causes stock-outs.
• Ignoring data: new parameters on inaccurate records give confident wrong answers.
• One-time projects: parameters drift back if nobody owns the quarterly review.

Where to go next

If your shortages come from internal processes rather than inventory rules, such as order errors, long changeovers or erratic production, the Lean Six Sigma and Operational Excellence courses address those causes. The Excel template from the Insights inventory series is a practical place to run your first ABC and safety-stock calculations.