A Class That Never Gets Written Back
Our ABC report ranks your catalogue by consumption value and bands it properly, on cumulative share rather than a crude top-fifth. Our counting module can hold a class against a count plan. Nothing carries the first into the second, so the analysis that should ration your counting effort is a PDF somebody retypes.
A catalogue of twenty thousand lines cannot be counted the same way twice a year. Some of those lines are the business and some of them are washers. The whole discipline of cycle counting rests on being able to say which, and then counting accordingly.
We can tell you which. We just cannot tell the counting module.
The half that works, and works well
The ABC report is better than most. It ranks every item by consumption value over a period you choose, then bands on cumulative share — class A is the set of items that together make up the first eighty per cent of your spend, however many items that turns out to be. That is the correct definition and it is not the common one. A great many systems band on row count, calling the top twenty per cent of lines "A" regardless of what they are worth, which produces a tidy-looking chart and the wrong answer.
Both cut-offs are adjustable, items with no consumption in the period are forced to C rather than being ranked as though they matter, and the output comes out as a spreadsheet or a PDF.
The counting module is also genuinely built. There are count plans and count sessions, assignments to specific people, a blind count that is actually enforced rather than merely recommended, per-line statuses including a recount request that carries a reason and writes an audit event, variance rules, and — the design decision that matters most — variances do not silently rewrite stock. They raise pending adjustments that somebody has to approve.
Two good pieces of work, built at different times, with no wire between them.
The wire that is missing
A count plan can hold an ABC class. The ABC report computes one. Nothing writes the second into the first.
So the sequence a warehouse manager actually performs is: run the report, export it, open it beside the count plan screen, and set the class by hand, item by item, for as many items as they have patience for. Then do it again next quarter, because consumption moves and the classification with it.
In a catalogue of two hundred lines that is an afternoon and nobody writes a blog post about it. In a catalogue of twenty thousand it is not a task, it is a reason the classification never gets refreshed — and a stale ABC classification is worse than none, because the counting schedule built on it now protects the wrong items with everybody's confidence behind it.
What a stale class costs
| What happened | What the plan still says | The consequence |
|---|---|---|
| A line became your biggest seller in March | C, from last year's run | Counted annually, so a discrepancy sits undetected for months on your highest-value movement |
| A line was discontinued in January | A, from last year's run | Counted monthly, at full effort, on stock nobody is moving |
| A new SKU launched in April | No class at all | Falls into whatever the default treatment is, which is nobody's decision |
| Seasonal lines rotate through the year | One fixed class | Half the catalogue is misclassified for half the year, by construction |
The last row is the general case, and it is the one that argues for the wire most strongly. Classification is not a property of an item; it is a property of an item at a point in time. A manual copy-across is not merely laborious, it is guaranteed to be out of date, because the thing being copied changes continuously and the copying happens occasionally.
How to run this today
A workable quarterly rhythm with the pieces that exist
- Run the ABC report on a rolling twelve months, quarterly, not on a calendar year. A rolling window smooths seasonality without hiding a trend, and quarterly is frequent enough that a class is never more than one quarter stale.
- Only maintain the A band by hand. This is the concession that makes the workaround survivable. The A items are, by the report's own definition, the ones making up eighty per cent of your spend, and there will be far fewer of them than you expect. Classify those properly and let B and C default.
- Watch the movement between runs, not the bands. The useful output of the quarterly run is not the new classification, it is the list of items that changed band. That list is short, and it is the only part anybody needs to act on.
- Use assignments to spread the load. The counting module lets you assign lines to named people. A weekly slice of the A band, assigned and blind, beats a quarterly heroic effort on everything.
This is a wire between two things that both already exist
It is worth being precise about how small this one is. Neither side needs building. The ranking is computed and correct; the field on the count plan exists and is stored. What is missing is the act of carrying one into the other, plus the decisions about when it should happen and what it should not overwrite.
Write the class back
A run of the ABC report stamping its result onto the count plan, with a record of when it was stamped and by which run.
A manual override that survives
The harder half. Some items are class A for reasons no consumption figure knows about — a regulated line, a customer commitment, a theft target. An override has to be respected by the next run rather than flattened by it.
A schedule that reads the class
Worth naming because it is a separate absence: a count plan's frequency is stored and nothing consumes it either, so a class-driven schedule needs the scheduler as well as the class.
We do not publish dates. If cycle counting at catalogue scale is central to your operation, tell us the shape of it and we will come back with a written scope, a timeline and a cost before you commit.
Scope the counting workflowWhat AWRA OpsHub does today
- ABC analysis banded on cumulative consumption value with both cut-offs adjustable, zero-consumption items forced to C, exported as CSV or PDF.
- An inventory turnover and days-inventory-outstanding report over any date range, with consumption summed from both stock issues and till sales.
- Count plans, count sessions and per-person assignments, with a genuinely enforced blind count.
- Per-line count statuses including a recount request carrying a reason, with audit events, and a session that cannot close while lines are unresolved.
- Variances raised as pending adjustments for approval rather than written straight to stock.
- Daily dead-stock scanning against a per-organisation idle threshold, and batch-expiry alerting on a 30/60/90 horizon.
What it does not do
- Any write-back of the ABC class onto the count plan. The report produces it, the plan can hold it, nothing carries it across.
- Scheduling from a count plan. The frequency field is stored and validated and nothing reads it, so no session is ever created automatically.
- A count accuracy measure. Variance is captured on every line and nothing aggregates it into an accuracy percentage, a trend, or a score by counter or location.
- Category or location roll-ups, trends or benchmarks on the turnover report.
- Per-item dead-stock thresholds — the idle period is one figure for the whole organisation.
Not ours, by choice
- The ABC banding is on cumulative value, which is the correct definition and not the common one. We would rather be judged on that than on the gap above it.
- The manual copy-across genuinely works at small catalogue sizes. It fails at scale not because it is hard but because it will not be repeated often enough.
- Nothing on this page is Polish. It is the arithmetic of a large catalogue; Poland is where our reader is most likely to have one.
Our position
The classification and the counting are both real and both good. The join between them is manual, and at catalogue scale a manual join is a join that stops happening. Run the report quarterly on a rolling year, maintain the A band by hand and only the A band, and act on the list of items that changed class rather than on the whole classification. If your catalogue runs to five figures and counting effort is your constraint, raise the write-back with us as a requirement rather than discovering it in month three.
Get your first ABC run read properly
The first run is the one worth interpreting with somebody. Most catalogues turn out to have a far shorter A band than their owners expect, and that single fact usually changes the counting plan more than any feature would.
Walk through your catalogueFrequently asked questions
Can I import the ABC classes in bulk instead of typing them?
There is no import that targets the class on a count plan, so the copy-across is manual on the screens. The practical mitigation is to maintain only the A band, which is far smaller than most people expect once the banding is on cumulative value rather than row count.
Does the ABC report use till sales as well as stock issues?
The turnover report does, deliberately — a shop that sells only through the till would otherwise report zero consumption. Treat that as the reference for how consumption is measured, and check the ABC output against your own expectation on a handful of known items before you build a plan on it.
If the count plan frequency does nothing, what actually triggers a count?
A person. Sessions are created by hand, and the plan's frequency is a record of intent rather than a schedule. That is worth knowing before you write a counting policy that assumes the system will prompt anybody.
