Cycle Count Planner

What this tool does

CM8-353 plans your cycle counting. It classifies every item by annual usage value, applies a count frequency to each class, projects the resulting workload week by week against the counting time you actually have, and tracks how accurate the counts turn out to be — separately for each class, because that is where the answer lives.

Everything runs inside this single file — no account, no upload, no network request of any kind.

Why cycle count instead of stocktake

An annual stocktake tells you how wrong your stock was, once, after it has already been wrong for a year. Cycle counting spreads the same effort across the year so errors are found weeks after they happen, while the cause is still findable — and because you count the valuable items often and the trivial ones rarely, you spend most of the effort where being wrong actually costs something.

The trap is building a plan that cannot be executed. A plan asking for eleven hours of counting a week from somebody who has five will not produce accurate stock; it will produce skipped counts, and the skipped ones will be the awkward items, which are exactly the inaccurate ones. That is the failure this tool is built to catch.

What one row is

One row is one item you count — not one count and not one location. An item held in two bins is still one row; note it, and count both together, because a count of one bin against a system balance covering both is not a variance, it is arithmetic nonsense. The sample includes one of these.

How classification works

Items are ranked by annual usage value, highest first. The tool then walks down the list adding up value as a share of the total:

This is the classic Pareto split, and in most stores it produces something close to: a fifth of the items carrying four-fifths of the value. If class A comes out as a third of your item list, either the cut is too generous or your usage really is that evenly spread — check the donut before changing anything.

Classification is relative, so it moves. Add a large new item and everything below it shifts down; remove one and things shift up. That is correct — class is a statement about an item's rank within your stock, not a property of the item — but it does mean an item can change class without anybody touching its row.

Usage value, not stock value

The figure that drives everything is annual usage value: how much of the item you get through in a year, multiplied by what it costs.

Not the value sitting on the shelf. The reason is that counting exists to catch errors, and errors are created by movement. An item that moves five thousand times a year has five thousand chances to be recorded wrong; an item sitting untouched in a cage has almost none, however much it is worth.

That gives the one case where the rule breaks, and the sample shows it: a critical spare with zero usage and enormous value. Classification would put it in C. Force it to A — not because it moves, but because "we thought we had one" is a sentence nobody wants to hear when the line is down.

When to override the class

Sparingly, and for a reason you write in the notes. Three good ones:

• Critical spares — high value, no usage, catastrophic if missing.
• Items with a compliance or traceability requirement — where counting frequency is set by somebody other than you.
• Freely issued consumables — technically valuable in aggregate, but nobody is going to control a glove bin, so counting it monthly is theatre. The sample forces one of these down to C.

Choosing frequencies

Twelve, four and one a year is the common starting point and a reasonable default. Some things worth knowing before you change it: doubling class A frequency roughly doubles your class A counting time but usually improves accuracy far less than doubling; and counting class C more than once or twice a year almost never repays the effort.

An item that has never been counted is due today, not in a month. That is deliberate: a never-counted item is the one most likely to be wrong.

The workload chart

Thirteen weeks forward, in hours, with your capacity as a dashed line. Overdue counts are not piled into week 1 — they are reported separately, because a first bar containing a year of arrears tells you nothing about whether next week's plan fits.

If the plan is over capacity, the honest options are the same four every time: count class C less often, make counting faster (labels, access, one location per item, a scanner), buy more counting time, or move the class A cut so fewer items are class A. What is not an option is leaving it and hoping, which is what happens by default.

Measuring accuracy honestly

Record two numbers per item: how many counts were done in the last year, and how many of those came out within tolerance.

Counting hits rather than averaging variances is deliberate. An average variance of zero can mean every count was perfect, or that half were 20% over and half 20% under — which is a store in chaos. Percentage-within-tolerance cannot hide that.

Judge accuracy by class, never overall. An overall figure is dominated by class C, where hundreds of low-value items are counted by weight and nobody seriously expects a match. Class A below target is the finding that matters, because it means the value on your balance sheet is wrong.

What to do with a bad count

A variance is a symptom. Adjusting the balance and moving on treats the symptom and guarantees you will be back next month. The causes worth looking for, in rough order of frequency: issues not booked out, receipts booked to the wrong item, unit-of-measure confusion (each versus box), stock in a second location nobody counted, and returns never booked back in.

The location chart exists for this. Overdue and inaccurate counts clustered in one place is telling you something about that place — poor access, no labelling, stock in two bins — not about the person counting.

What this does not do

• It does not count anything, hold stock balances, or post adjustments.
• It does not know what your system says you have. Variance is a number you type in after a count.
• It does not schedule counters or produce count sheets. The "what to count next" table is the list; who does it and when is yours.
• It assumes a count takes the same time every time, which is close enough for planning and wrong for any individual count.
• It has no view about whether your tolerance is reasonable. Two percent on a bearing and two percent on a bin of washers are very different promises.

Printing and sharing

The Report tab prints the tiles, charts and both tables with a title block you fill in. Filter to a location before printing if you want a count list for one area.

Saving your work

The plan is held in this browser, on this computer, and stays there between visits. Use the backup button to write a JSON file you control; the spreadsheet download gives you the same rows, which is the easiest way to get a count list into somebody's hands.

Accuracy & disclaimer

Every figure here is yours: usage values, count times, variances and hits. The tool ranks, schedules and adds up. It cannot tell you whether a count was done properly, and an accuracy figure built from counts nobody supervised is a measure of paperwork rather than of stock.