ABC Inventory Analysis: How Do You Grade SKUs Without Hiding Risk?

ABC inventory analysis ranks products by their contribution to a chosen value measure, then groups them so your team knows where to spend attention. For an ecommerce store, the useful output is a short action list: which SKUs need availability checks, which deserve investigation, and which can tolerate simpler controls.

An A grade is not permission to buy more. A product can lead revenue while generating weak contribution or suffering expensive returns. Start with the grading worksheet below, then add those exceptions before changing inventory policy.

Table of contents

• What does ABC analysis measure?
• How do you build the worksheet?
• How do you handle the threshold boundary?
• What can make an A grade misleading?
• What actions belong to each class?
• How do you keep the grades useful?

What does ABC analysis measure?

ABC measures relative importance within the population you include. It does not measure whether a product is profitable, whether the next purchase order is sensible, or whether its supplier is reliable.

Choose the value measure first. A revenue analysis answers which products carry sales; an annual consumption-value analysis answers where purchased inventory value is concentrated. Keep separate worksheets if you need both. Multiplying units by selling price and multiplying units by unit cost answer different questions.

Shopify's ABC guidance describes approximate revenue shares of 80% for A, 15% for B, and 5% for C. These are shares of value, not a requirement that exactly 20% of SKUs become A. Its documented revenue measure excludes discounts and does not use product cost, so it should not be relabeled a profit ranking.

For a small catalog, one expensive item can move the threshold sharply. That is a property of the assortment, not a spreadsheet error. Keep the actual class shares visible beside the target shares.

How do you build the worksheet?

Build one row per sellable SKU, use one reporting window, and rank the chosen value from highest to lowest. Preserve enough source detail that another person can reproduce every total.

Use these columns: SKU, product name, period, units, value, share of total, cumulative share, class, exception, owner, and next action. Record whether refunds, discounts, taxes, shipping income, and currency conversions are included. Do not combine differently defined revenue exports.

Here is an illustrative six-SKU store, not SellerTrove customer data:

The arithmetic is revenue divided by $10,000; cumulative share is the running total. For a platform report, confirm its window rather than assuming it matches yours: Shopify's inventory-report documentation describes ABC grades using the last 28 days.

Before grading, reconcile the worksheet total with the same filtered source report. If it fails, investigate duplicate variants, excluded channels, currencies, and returns. Do not improve the formatting of an unreconciled export.

How do you handle the threshold boundary?

Choose a boundary rule before looking at the result, and apply it consistently. A SKU is indivisible for this decision, so actual class shares will rarely land exactly on 80%, 95%, and 100%.

Our worksheet assigns a row from the cumulative share before that row: below 80% means A; from 80% to below 95% means B; 95% or above means C. The item that crosses a boundary stays in the earlier class. This is an explicit worksheet convention, not a claim about every platform's implementation.

In the example, the lunch box starts at 70%, so it remains A and carries A to 85%. The brush starts at 93%, so it remains B. The resulting value shares are 85%, 13%, and 2%. Using the documented 80/15/5 convention as a target does not make these results incorrect.

For equal values, keep a stable secondary sort such as SKU. Alternatively, keep tied items together and accept a larger boundary difference. Record that decision so a harmless re-sort does not create apparent product movement.

What can make an A grade misleading?

The ranking is backward-looking and only as representative as its data window. Stockouts, promotions, new launches, and unusual returns can all make a clean-looking grade misleading.

Add exception flags without silently changing the underlying ranking:

• Unavailable: sales were limited by missing stock; low revenue does not prove low demand.
• Promoted: a temporary campaign changed the normal mix.
• New: the SKU has too little history for a stable classification.
• Low contribution: revenue is high but costs or returns absorb it.
• Dependency: a low-revenue spare part keeps a high-value product useful.

For example, the $800 replacement lid may protect the customer experience for the $4,500 mug. Calling it B does not justify allowing it to disappear. That dependency is an operating decision layered over the arithmetic.

Older tutorials may describe Stocky. Shopify's Stocky documentation states that the app was no longer available after August 31, 2026. Do not build a new October 2026 workflow around that retired interface; use the current reporting and inventory tools available to your store.

What actions belong to each class?

Use class to allocate attention, then let operational risk override the default. The goal is fewer consequential mistakes, not equal effort across every SKU.

Give each action a named owner and a due date. “Check availability” should mean reviewing stock, inbound commitments, and expected demand; it should not become an automatic purchase order. The grading method does not replace a forecast or reorder calculation.

Use the cycle-counting guide to turn risk into counting controls. Use the SKU rationalization scorecard when the decision is whether to keep, fix, bundle, or retire an item. These are separate decisions; an ABC export is useful input to both.

How do you keep the grades useful?

Refresh the worksheet on a schedule that matches your buying cycle, and explain meaningful movement before taking action. Keep the previous version so a changed grade has context.

A practical review records five things: old grade, new grade, value change, cause, and action. Distinguish growth in the product's own value from a higher share caused by other products declining. Both can promote an item to A; only one demonstrates its own growth.

Test the workflow on a small assortment before automating it. Ask a colleague to reproduce the $10,000 example and boundary decisions. Then run the same checks on your export: complete population, consistent currency, reconciled totals, stable ties, and documented exceptions.

When repeated exports become fragile, browse inventory tools for the actual capabilities you need: variant-level reporting, traceable adjustments, channel coverage, and dependable exports. Use the Stack Builder to organize related work. A tool that prints colored grades but cannot explain the numbers has not solved the job.