Inventory Management Techniques: Pick Per Item, Not Per Business
Inventory management techniques work best matched to each item: classify stock by value (ABC analysis) and demand pattern, give each class a reorder and counting method, and set valuation once by policy.
By Praveen Nune, Co-Founder & CEO · Updated 10 Nov 2026
Key takeaways
- Use no single technique for the whole catalogue. Rank items by annual value (ABC), grade them by demand steadiness (XYZ), and give each cell its own reorder and counting method; set valuation once by accounting policy.
- Your ABC cut-off decides your priority list. In one retail dataset, the same sales produced a class A of very different sizes at a 70%, 80% and 90% cut-off.
- Classes can move. In one retail dataset only 59% of year-one A items were A again a year later, while a steadier range kept 97.8%. Our recommendation: re-run the ranking quarterly if your range turns over.
- Under both Ind AS 2 and AS 2, inventory cost is assigned by FIFO or weighted average, and LIFO is not allowed.
In this article
- 1Which inventory management techniques should you use, and for which items?
- 2How do you build the ABC–XYZ picker from your own sales export?
- 3Where should you cut ABC analysis: 70%, 80% or 90%?
- 4How often do ABC classes change?
- 5Why does ranking by revenue alone give the wrong priority list?
- 6How do FIFO and weighted average fit with picking and counting?
- 7How do we help you run these techniques in Arka Inventory?
- 8Frequently asked questions
Most lists of inventory management techniques give you ten definitions and leave the real decision untouched: which technique belongs to which product. A business with a few thousand items cannot apply every technique to all of them. The effort has to go where the money and the risk are.
That decision rests on two things you can calculate from your own sales export in an afternoon: how much value an item carries, and how steadily it sells. Both are less stable than textbooks suggest. Where you draw the line for your priority items changes how many items count as priority by a wide margin, and the list itself shifts from one year to the next.
This page gives you the picker we use to assign a technique to each item, the numbers from two public sales datasets that show why the cut-offs matter, and the Indian accounting rule that keeps stock handling and stock valuation apart.
Which inventory management techniques should you use, and for which items?
Rank every item by annual value into A, B and C, then grade its demand as X, Y or Z. The cell it lands in tells you which reorder and counting technique to use. Valuation is set once for the business by accounting policy, not by cell.
Terms used in the table:
- Reorder point: daily demand × supplier lead time in days + safety stock. For example (illustrative numbers), 12 units a day × 9 days + 36 units of safety stock = 144 units.
- Safety stock: extra units held against demand spikes or late deliveries.
- Min/max: when stock falls to the minimum, order back up to the maximum.
- Periodic review: check on a fixed day and top up to a target level.
- Cycle count: counting a rotating subset of stock on a schedule instead of a full stocktake.
- Just-in-time (JIT): receiving stock only as it is needed.
- Vendor-managed inventory: the supplier monitors and refills your stock.
- Dead stock: items with no sales for a long period.
This table is our recommendation, built on the working in the sections below.
| Cell | Reorder method | Cycle count frequency |
|---|---|---|
| AX (high value, steady) | Reorder point; candidate for just-in-time (JIT) delivery or supplier-held stock (vendor-managed inventory) | Monthly |
| AY (high value, variable) | Reorder point with a larger safety stock | Monthly |
| AZ (high value, lumpy) | No average-demand reorder point. Order against confirmed orders and a forecast; review weekly | Monthly |
| BX, BY | Min/max levels | Quarterly |
| BZ, CZ | Make or buy to order, or stop stocking; review as possible dead stock | Quarterly to yearly |
| CX, CY | Periodic review and bulk orders | Once a year |
For order sizes, test your quantity against supplier minimums using our guide to economic order quantity, the order size that minimises ordering plus holding cost. For lumpy items, set targets using inventory forecasting methods.
How do you build the ABC–XYZ picker from your own sales export?
Export twelve months of sales by item, sort by value, mark the cumulative share, and add one demand-steadiness figure. It takes five steps:
- ExportOne row per item (stock keeping unit, or SKU): annual value, units, and the number of order lines. Manufacturers should use issue value (quantity × cost).
- SortBy value, largest first, and compute the cumulative share of total value.
- Assign ABCClass A runs until the cumulative share reaches your cut-off, B to the next cut-off, C is the rest.
- Grade demandSplit the year into thirteen 28-day periods, total the units sold in each, and divide the standard deviation by the mean. This ratio is the coefficient of variation (CV). X is 0.5 or below, Y is 0.5 to 1.0, Z is above 1.0.
- Override by propertyExpiry-dated, batch-tracked or serialised items get their own handling whatever their cell says.
The X/Y/Z breaks at 0.5 and 1.0 are our working convention, not a standard. Adjust them if too many items land in Z.
Where should you cut ABC analysis: 70%, 80% or 90%?
Cut where your class A list matches the number of items your team can count every month. The cut-off is a choice, and it changes the list size sharply.
We classified the Online Retail II dataset from the UCI Machine Learning Repository, which records a UK online gift seller selling mostly to wholesalers. We used 10 December 2009 to 10 December 2010 as Year 1 and 10 December 2010 to 10 December 2011 as Year 2. We removed cancellations, zero or negative quantities and prices, and non-product codes (postage, bank charges and similar). That left 4,167 SKUs in Year 1 and 3,857 in Year 2. Share of SKUs in class A:
| Cut-off (A/B/C share of revenue) | Year 1 | Year 2 |
|---|---|---|
| 70/20/10 | 14.5% | 14.1% |
| 80/15/5 | 21.9% | 21.2% |
| 90/7/3 | 35.1% | 34.3% |
The "top 20% of products make 80% of sales" rule is only a rough guide. In this dataset the top 20% of SKUs made 77.9% (Year 1) and 78.6% (Year 2) of revenue. In a second dataset, weekly unit sales of 811 products over 52 weeks, the top 20% made 66.6% of units. Measure your own catalogue before you quote the rule to your team.
How often do ABC classes change?
Re-run the ranking every quarter, because class A membership turns over, though how much depends on your range. Of the 606 SKUs in class A in Year 1 of the retail data (70/20/10 cut-off), 359 (59%) were A again in Year 2. Another 170 became B, 46 became C and 31 sold nothing. In Year 2, 136 of the 542 A items had had no sales in Year 1.
Within a single year the movement was smaller but real. Comparing the first six 28-day periods of Year 2 with the last six, 66.5% of the 466 first-half A items stayed A, and 7.5% fell to C or stopped selling.
The weekly-units dataset behaved differently. Comparing its first 26 weeks with the last 26, 174 of 178 A products (97.8%) stayed A, and none fell to C. A stable range with steady demand barely reshuffles; a seasonal, churning range reshuffles a lot. If your range turns over, as a wholesaler's or distributor's often does, a yearly ABC run goes stale before the year ends.
Why does ranking by revenue alone give the wrong priority list?
Revenue ranking misses two things: workload and demand shape. Both change what you should do with an item.
Workload. In Year 2 of the retail data, the top 20% of SKUs by revenue (771 SKUs) overlapped the top 20% by units by only 65%, and the top 20% by order lines (the number of times an item is picked) by 79%. Those revenue leaders made 78.6% of revenue but only 60% of order lines. If you slot a warehouse or schedule counts, rank by order lines as well as value.
Demand shape. We graded the 541 class A SKUs (70/20/10) on unit demand in thirteen 28-day periods of Year 2:
- 149 (27.5%) were X, 232 (42.9%) were Y and 160 (29.6%) were Z.
- 184 (34.0%) had at least one 28-day period with no sales.
- The median CV was 0.68 for A items and 1.52 for C items.
One item (a class A SKU that sold only on the final day of the year) is left out of this grading, which is why the count is 541, not 542.
Our inference: an "average daily demand" reorder point is safe for X items and risky for Z items. For a Z item, the average hides long gaps and sudden orders. That is why the table above sends AZ items to order-driven planning.
How do FIFO and weighted average fit with picking and counting?
Keep stock rotation and stock valuation as separate decisions. FIFO as a picking rule (oldest stock leaves first) protects expiry-dated goods. Valuation is an accounting choice. Under both Ind AS 2 (Inventories) and AS 2 (Valuation of Inventories), inventory cost is assigned by FIFO or weighted average, and LIFO (last in, first out) is not allowed.
AS 2, issued by the Institute of Chartered Accountants of India (ICAI, revised 1999, mandatory for periods beginning on or after 1 April 1999), applies to companies not covered by Ind AS. It says cost "should be assigned by using the first-in, first-out (FIFO), or weighted average cost formula." Ind AS 2 applies to companies that must follow Ind AS. Your auditor can confirm which of the two applies to your company.
You can pick the newest stock for physical convenience and still value by FIFO or weighted average. For the choice between the two formulas, see our guide to inventory valuation methods.
How do we help you run these techniques in Arka Inventory?
We build Arka Inventory, cloud inventory software native to Salesforce, to carry out the handling side of the picker. Run the ABC–XYZ ranking from your sales export; then we help you act on it:
- Reorder follow-through. Inventory alerts flag stock approaching or falling below minimum levels and back orders arriving. This is on our Basic plan.
- Stock rotation. We pick by FIFO for non-lot items and pick lots with the earliest expiry first. See our warehouse management features.
- Batch, expiry and serial tracking. For items your override step marks, lot and serial tracking is on our Enterprise plan.
- Multi-warehouse visibility and purchase planning. We show stock by location, stock on order and lead times, and plan purchases from real orders, forecasts and stock. Multiple warehouses and purchase orders are on Basic. Sales orders are on Enterprise. For forecasting and purchase automation, see plans and pricing. Bills of material, production and kitting are added on our Advance plan.
Start your 15-day free trial: full access to all features, no credit card needed. Onboarding typically takes at least two weeks, depending on how ready your item, warehouse and stock data are. You can also see plans and pricing, ask for a demo, email support@arkainventory.com, or call +91 93902 73833 (Monday to Friday, 9 AM to 6 PM Indian Standard Time).
Frequently asked questions
Sources
- 1.Accounting Standard (AS) 2, Valuation of Inventories, revised 1999 (AS 2, non-Ind AS companies), Institute of Chartered Accountants of India
- 2.Ind AS 2, Inventories, Institute of Chartered Accountants of India
- 3.Online Retail II dataset (donor Daqing Chen, donated 20 September 2019, CC BY 4.0). Our ABC, migration, revenue-share, overlap and CV results were computed from this data on 10 November 2026 using the cleaning and windows described above., UCI Machine Learning Repository
- 4.Sales Transactions Dataset Weekly (811 products, 52 weeks of unit sales). Our top-20% share and half-year class-stability results were computed from this data on 10 November 2026., UCI Machine Learning Repository
- 5.Arka Inventory: warehouse management features
- 6.Arka Inventory: pricing
- 7.Arka Inventory: contact sales
