Stop Shutting Down for Stocktakes: Cycle Counting, A Better Way to Manage Inventory

The annual shutdown stocktake is a ritual dressed up as a process. There is a better way, and it doesn't require locking your doors.

A perspective on cycle counting for small and mid-size manufacturers and distributors

Here is something worth sitting with.

Every year, thousands of manufacturing and distribution businesses grind to a halt. They stop shipping. They stop producing. They call all hands. And they count.

It takes days. It costs a fortune in lost output and overtime. And at the end of it, the number they get is already slightly wrong, because goods moved the moment the count ended.

The shutdown stocktake isn't a solution. It's a retreat.

It's a signal that the business doesn't trust its own data enough to rely on it day to day. So once a year, it creates a moment of forced truth. Painful, expensive, and exhausting.

The alternative has a name: cycle counting. And if you run a manufacturing or distribution business of any meaningful size, it will change how you think about inventory — not just how you count it.

Counting everything at once doesn't give you accuracy. It gives you a snapshot of one chaotic moment. Counting a little, always, gives you something far more valuable: a system that knows itself.

What Cycle Counting Actually Is

Instead of one enormous count, you divide your inventory into sections and count a rotating portion continuously — daily, weekly, or monthly, depending on how critical those items are. High-value components get counted more often. Slow-moving consumables get counted less.

The warehouse doesn't stop. Production doesn't pause. Shipments keep moving.

But discrepancies get caught in days, not twelve months from now when the damage has already compounded. The system stays honest. The people running it stay sharp.

It sounds simple because it is. But simple isn't the same as easy. For cycle counting to work — genuinely work, not just exist on paper — there are non-negotiable requirements. Let's walk through each one, what it demands, and what it looks like in a real manufacturing or distribution setting.

The Requirements for Making It Real

Requirement 1 - Your Baseline Data Has to Be Worth Counting Against

Cycle counting compares physical reality to your system's record. If your system record is fiction, all you're doing is repeatedly confirming that fiction is wrong. That's not progress.

Before you start, you need a system of record — an ERP, WMS, or even a disciplined spreadsheet — that is updated in real time for every receipt, pick, transfer, and return. Not batch-updated at the end of the day. In real time.

Manufacturing Example
A mid-size metal fabrication business starts cycle counting and finds a 15% variance on steel tube offcuts. The root cause? Operators had been logging material usage at the end of their shift from memory, not at the point of cut. The system thought they had more than they did. The fix wasn't better counting — it was a simple scan point at the cutting station.

The Workflow

1. Audit your last three months of inventory adjustments. If adjustments are frequent and large, your data entry process is the problem — fix that first.
2. Identify every point where stock moves and confirm there is a real-time system transaction for it: goods-in, production issuance, finished goods, despatch, returns, and scrap.
3. Set a threshold: if your current system accuracy is below 85%, do a one-time full count to reset the baseline before beginning cycle counting.

Requirement 2 - Every Item Needs a Home and That Home Needs a Label

Cycle counting without fixed locations is like trying to audit a library where the books can be shelved anywhere. You might find them. You might not. Either way, you can't be sure you've found them all.

Every SKU needs a designated bin, shelf, rack, or zone. Every location needs a clear, scannable label. The system needs to know what's supposed to be where, so counters know exactly where to look — and where to record.

Distribution Example
A parts distributor with 6,000 SKUs was counting the same items twice — once in their primary pick location, and once in the overflow reserve bay — without reconciling them. Their location field in the system was optional. Making it mandatory, and assigning formal codes to every bin (A-01-03 = Aisle A, Shelf 1, Position 3), cut their phantom variance by 40% in the first month.

The Workflow

1. Map your physical storage. Assign alphanumeric codes to every distinct location: aisle, bay, shelf, and bin level.
2. Print and laminate location labels. Apply them physically. This sounds obvious — it's consistently the step businesses skip.
3. Update your inventory system so every stock record carries a primary location code. No location, no record goes live.
4. Establish a put-away rule: stock can only be placed in its designated location unless a system override is raised and documented.

3. A Schedule That Runs on Logic, Not Guesswork

Cycle counting without a schedule is just random spot-checking. You get the psychological comfort of counting without the systemic benefit of coverage.

The schedule should be driven by ABC analysis — a classification of your inventory by value and velocity. High-value, high-movement items (A) get counted monthly. Mid-range items (B) quarterly. Low-value, slow-moving items (C) twice a year. Your riskiest stock gets the most scrutiny, automatically.

Manufacturing Example
A food packaging manufacturer classified its 1,200 raw material SKUs. The top 80 items — specialist adhesives, printed film, and aluminium foil rolls — represented 78% of raw material cost. Counting just those 80 items monthly, spread across four short counts per week, added under 90 minutes of counting time weekly. It caught three significant discrepancies in the first quarter that had been silently accumulating for years.

The Workflow

1. Pull a spend or usage report for all SKUs. Rank by annual value (unit cost × annual consumption).
2. Classify the top 10–20% by value as A-items, the next 30% as B, the remainder as C.
3. Build a counting calendar. Assign specific locations or item groups to specific days. Protect this time in the shift schedule.
4. Review the ABC classification every six months — products change, so should the schedule.

Requirement 4 - Counting During Live Operations Requires Traffic Control

This is the requirement most businesses underestimate. If stock is being picked from a location while someone is counting it, you will get a variance — not because your data is wrong, but because the count and the movement happened simultaneously.

You don't need to freeze the whole operation. You need to freeze the specific location being counted for the duration of that count. That's a matter of minutes, not hours. But it requires coordination.

Distribution Example
A wholesale electrical distributor implemented a simple count flag in their WMS. When a location was scheduled for a cycle count, the system automatically placed a 20-minute picking hold on it. Pickers were routed to alternative locations. The counter completed the count, submitted it, and the hold released automatically. No manual coordination required. Variance from in-progress picks dropped to near zero.

The Workflow

1. Define your counting window for each location — ideally a period of naturally lower activity such as shift start or after the main despatch run.
2. Either use a system hold on the location or communicate to the floor supervisor that the location is temporarily frozen.
3. Where system holds aren't available, use a visible physical marker — a coloured tag or cone — that signals to pickers to hold off until removed.
4. Log the start and end time of each count. Any system transactions within that window for that location must be reviewed against the count result.

Requirement 5 - Counters Must Count Blind

This is the one that gets argued about most. It is non-negotiable.

If a counter can see what the system says should be there before they count, they will — consciously or not — count until they find that number. It's human nature, not dishonesty. The result is that your count confirms your system rather than challenges it.

Blind counting means the counter records what they physically find. The system compares that figure to the expected quantity after submission. Discrepancies are then surfaced and investigated.

Manufacturing Example
An automotive components manufacturer gave counters sheets showing item descriptions and locations — but not system quantities. When counts were submitted, the system flagged variances automatically. In the first six weeks they found a consistent 12-unit shortfall on a specific fastener. Investigation revealed a supplier delivering bags of 488 instead of the stated 500. Twelve fasteners per delivery, undetected for over a year. Blind counting surfaced it. Non-blind counting had buried it.

The Workflow

1. Configure your inventory system to produce count sheets or mobile scanner tasks that display item description and location — but suppress the system quantity until after submission.
2. If your system doesn't support this, use a two-stage paper process: count sheet submitted first, then compared against a separately printed expected-quantity report by a supervisor.
3. Train counters explicitly on why blind counting matters. They are not auditors looking for problems — they are the accuracy system doing its job.
4. For significant variances, assign a second independent counter to recount before escalating to a full investigation.

Requirement 6 - A Variance Isn't the End, It's the Beginning

Finding a variance is not a failure of cycle counting. It is cycle counting working exactly as intended. The failure is treating every variance as noise to be adjusted away rather than a signal to be understood.

Every unexplained variance above your tolerance threshold needs a root cause category: counting error, process failure, system error, damage, shrinkage, supplier discrepancy. Over time those categories tell you where your operation is leaking — and that is genuinely valuable intelligence.

Distribution Example
A building materials distributor categorised six months of variance data and discovered that 62% of their negative variances fell into a single category: unprocessed returns. Goods were coming back from job sites and being shelved without a system return being raised. Nobody was stealing anything. Nobody was miscounting. There was simply a gap in the returns process. One procedural change — returns held in a dedicated bay until system-processed — resolved the majority of their ongoing variance problem.

The Workflow

1. Set a tolerance policy: variances within a defined percentage or below a set dollar value are auto-adjusted with a reason code. Variances above the threshold require investigation before adjustment.
2. Assign reason codes to every variance: counting error, system entry error, process gap, damage or scrap, supplier short-ship, shrinkage.
3. Review the reason code distribution monthly. If the same code dominates, you have a process problem — not a counting problem.
4. Require management authorisation for all inventory adjustments above a set value. Adjustments without oversight quietly undermine the integrity of the whole system.

Requirement 7 - Leadership Has to Actually Want This to Work

Cycle counting is easy to commit to in a slow week. It is the first thing cut when a big order lands, a machine breaks down, or the team is short-staffed.

That's the trap. The weeks when pressure is highest are the weeks when inventory accuracy matters most — when you need to know exactly what you have to make a promise to a customer, plan a production run, or avoid an emergency purchase at a premium price.

Leaders who protect counting time, act on variance findings, and treat inventory accuracy as a genuine business metric — those businesses build systems that stay reliable. Everyone else is going through the motions until the next painful annual shutdown.

Manufacturing & Distribution Example
The businesses that do this well share one visible behaviour: the operations manager reviews the weekly variance report in the same meeting as output, despatch, and quality. Inventory accuracy sits alongside the metrics that drive the business. It is not a back-office exercise. It is treated like the real asset it is.

The Workflow

1. Set a target inventory accuracy rate — 95% or above by location is a reasonable starting goal for most businesses.
2. Report accuracy weekly alongside other operational KPIs. Make it visible at every level of the operation.
3. Protect the counting schedule. Treat it as non-negotiable as a production schedule or despatch run.
4. Celebrate when variances lead to process improvements, not just when the numbers look clean. The finding is the win.

The Honest Truth About the Transition

Moving from annual shutdown counting to cycle counting feels harder than it is. The hardest part is the first few weeks, when count results reveal just how much uncertainty had been hiding in the system. That discomfort is the process working.

Within three to six months, something shifts. The team trusts the numbers. Purchasing decisions get sharper. Customer promises get more confident. Production planning gets cleaner. And you stop losing one to three days of output per year to a counting ritual that was solving the wrong problem.

You don't need a large technology investment to start. You need a location structure, a schedule, a discipline around blind counting, and someone willing to look at variance data each week and ask: what is this telling us?

Where to Begin

Pick your ten most valuable SKUs. Assign them fixed locations. Count them this week — blind, without looking at what the system says. Submit the result. Look at the variance.

That is your first cycle count. It costs nothing but an hour of someone's time. And it will tell you more about your inventory system than any annual shutdown ever has.

The question isn't whether you can afford to implement cycle counting. It's whether you can afford to keep not knowing what you actually have.

Inventory Operations Series  ·  Manufacturing & Distribution Edition