Cycle Counting Without Shutdowns: A Scheduling Playbook

One of our clients, a home goods e-commerce brand, ran a full physical inventory last December. They shut down shipping for two full days. The owner camped out in the warehouse office while sixty-plus people counted from morning to midnight. The result? A 3.1% book-to-physical variance — enough missing value to buy two forklifts. Worse, three weeks after the count, customer service started getting "out of stock" complaints again. The numbers they had just "fixed" were already drifting.

Afterward the owner asked me: your 3PL warehouses ship every single day — how do you do cycle counting? Honest answer: it's not about headcount. It's about scheduling.

I've signed over a hundred warehousing contracts, and the inventory accuracy clause is one I never let slide at the negotiating table. Here's the full playbook: how cycle counting stacks up against the annual wall-to-wall count, how to set ABC frequencies, how to schedule counts without stopping operations, and how to write the SLA in your contract.

Do the Math on the Annual Count — It Doesn't Pencil Out

A lot of operators swear by the big annual count. "We touch every location once a year, so we know where we stand." Run the numbers and that confidence falls apart.

Here's the counterintuitive part: the once-a-year "accuracy" is an illusion. The numbers look their best on count day, but that figure is the endpoint of twelve months of accumulated errors, not the starting point of a clean year. Product keeps moving, people keep mis-picking, and accuracy starts decaying the morning after. Cycle counting touches every location multiple times a year, so a variance gets caught days after it appears — while the root cause is still warm enough to find.

An auto parts client of ours used to do two full counts a year and hovered around 96% book accuracy. Eight months after switching to cycle counting, they stabilized above 99%. They actually saved labor — no more borrowing staff from other sites for count week, just one dedicated counter on a fixed daily routine.

ABC Segmentation: Spend Your Counting Hours Where They Matter

Cycle counting is not "grab a few random locations each day and hope." That is gambling. The professional approach is ABC segmentation:

• A items: high velocity, high value SKUs — counted weekly. Over a year, each A location gets counted 50+ times.
• B items: medium movers — counted monthly.
• C items: slow movers and dead stock — counted quarterly, which doubles as a dead-stock review.

Don't set the tiers by gut feel. Pull three months of outbound data from your WMS, rank SKUs by units shipped and value, and the A tier will reveal itself: a small slice of SKUs carrying the bulk of your shipments. That's what you watch daily.

Time for some arithmetic. Take a 100,000 sq ft warehouse with 8,000 locations, of which 800 hold A items. A counter scans and verifies 30–40 locations per hour, so full A coverage takes 20–27 labor hours per week — spread over five days, that's 4–5 hours a day. One dedicated counter handles it. B and C tiers ride along on the same routine.

Location barcodes are the lifeblood of cycle counting. No barcodes — or barcodes that peel off and never get replaced — and your counters are identifying product by sight. Sight means mistakes, and mistakes mean the count was theater. When we onboard a new client, re-labeling the entire building is job one. Don't cheap out on it.

Three Rules for Counting Without Stopping the Line

The enemy of in-operation counting is movement: you're counting a slot while someone else is putting product into it. Three rules fix that.

Rule one: slice by aisle, freeze one slice at a time. Divide the warehouse into zones by aisle or area and count one zone per day. In the WMS, freeze those locations during the count — no putaway, no picking, no moves. Unfreeze when the zone clears. Business barely notices. Just don't get greedy: freeze half the building and the pickers will revolt.

Rule two: count in the troughs. Schedule counts in the gaps between outbound waves — say, the quiet hour or two after the morning wave ships, or the start of night shift before picking ramps up. One client locks A-item counting to 6:00–8:00 AM daily; pickers clock in at 8:30. Zero overlap.

Rule three: count sheets are born and die the same day. The classic failure mode is a count sheet issued on Monday that someone works through until Thursday. Product moved a hundred times in between, and every variance multiplies. Our rule: tasks drop from the WMS in the morning and must close the same day; anything open at end of shift escalates to a supervisor. Counting delayed is counting corrupted.

When you find a variance, don't adjust it away yet. Run this five-point root-cause checklist first — it catches the real culprit most of the time:

1. Pull every inbound/outbound transaction for that location and SKU over the last three days — most "losses" are missed or wrong scans;
2. Check adjacent locations — product in the wrong slot is the most common form of "missing";
3. Check the returns and damage areas — returned goods that never got put back show as on-hand in the system but aren't in their slot;
4. Check in-transit picks — product picked but not yet packed may already be deducted in the system;
5. Check unit-of-measure conversions — a case of 12 recorded as 1 each creates a phantom variance of 11.

Adjust only after the investigation, and log every adjustment: who, why, and the evidence. In a 3PL warehouse, adjustment logs are shown to the client. No log, no credibility.

The Contract Clause: Three Questions for Your Next Negotiation

This is my home turf. The inventory accuracy SLA belongs in every 3PL contract, but most shippers negotiate it sloppily — a single line like "Provider shall maintain accurate inventory." That sentence is worthless. Accurate how? Measured when? And what happens when it isn't?

Ask these three questions next time you negotiate:

Question one: what is the denominator? Is accuracy measured per SKU line, per location, or per unit? The three methods can differ by several percentage points. In contracts I've signed, the common formulation is "monthly count accuracy of 99.5% or higher" — the exact figure is set at the table and the final contract governs. Nail down the basis in writing; I recommend per counted location, with spot counts and full counts specified separately.

Question two: who pays for the variance? When product is lost, does the 3PL reimburse at cost or at retail? Is there a deductible? The common structure is a tolerance band: the provider absorbs shrinkage up to an agreed percentage of inventory value per month, and anything beyond that is negotiated case by case. Set the percentage and the valuation basis at the table and put them in an exhibit — never leave them verbal.

Question three: who pays for the counting itself? If an annual count requires a shutdown, who eats the lost shipments? Who pays for temp labor? Is daily cycle-count labor baked into the storage fee? Leave this vague and you'll find a line item called "inventory count services" on the December invoice — and you'll have a hard time refusing it.

One piece of blunt advice from the provider's side of the table: the more the shipper understands inventory accuracy, the less the 3PL dares to fudge it. When the SLA basis, the variance handling, and the cost allocation are all spelled out, the warehouse will run its cycle count schedule properly — because it knows someone is watching that number every month.

By Susan, 3PL Sales Director