Cycle Counting Best Practices: How to Achieve 95%+ Inventory Accuracy

Inventory accuracy below 95% means stockouts, phantom stock, and production delays. Most manufacturers experience this because they rely on annual physical counts — a disruptive, once-a-year snapshot that's out of date the moment it's done. Cycle counting solves this.

I've helped 30+ manufacturing teams transition from annual physical counts to rolling cycle count programs. The results are consistent: accuracy improves within the first month, shrinkage becomes visible for the first time, and the team stops dreading inventory day because there is no inventory day anymore.

What Is Cycle Counting?

Cycle counting is a rolling count of a subset of your inventory on a regular schedule, rather than counting everything at once. Instead of shutting down operations once a year to count every item, you count a portion of inventory continuously — daily, weekly, or monthly — so that every item gets counted multiple times per year.

The result: errors are caught quickly, discrepancies are investigated before they compound, and your system records stay accurate year-round.

Cycle Count vs. Physical Count

Annual physical counts have three major problems:

• They require shutting down. Production halts, staff are pulled from their roles, and the business loses a day (or more) of productivity.
• Accuracy decays immediately. As soon as the count is done, transactions start creating discrepancies. By month six, your numbers may be significantly off.
• Errors aren't caught until it's too late. If a counting mistake or system error goes undetected for six months, it has downstream effects across procurement, production, and financials.

Cycle counting is continuous and real-time. Errors are caught within days or weeks — not six months later. And because counts are smaller and more frequent, staff can perform them without disrupting normal operations.

One thing I've noticed: manufacturers who resist switching from annual counts usually say the disruption isn't worth it. But they're comparing the pain of the annual count to an imagined ideal — not to the actual cost of running six months on inaccurate numbers. Phantom stock that triggers a false "we have enough" signal is expensive. You just don't see the bill until a production run comes up short.

ABC Analysis for Cycle Counting

Not all inventory deserves the same counting frequency. ABC analysis categorizes items by value and usage velocity so you can focus effort where it matters most:

• A items (high value or high velocity) — counted weekly. These represent your top 10-20% of SKUs but often 70-80% of your inventory value. Any error here has an immediate financial impact.
• B items (moderate value or velocity) — counted monthly. These are the middle tier: important but not critical enough to count every week.
• C items (low value or low velocity) — counted quarterly. These items rarely move and represent minimal financial risk.

In Nstock, ABC classification is automatic based on your actual usage patterns. The system identifies which items belong in each tier and surfaces them in your cycle count schedule without manual configuration.

For a manufacturer with 200 SKUs, you're typically looking at 30–40 A items, 80–100 B items, and the rest in C. That means your weekly count workload is 30–40 items — 20 to 30 minutes of focused work, not a half-day shutdown.

How to Run a Cycle Count: Step by Step

Step 1: Select items to count

Pull the list of items scheduled for counting this cycle. In a digital system, this is automatically generated based on your ABC tier and last count date.

Step 2: Print or pull the count sheet

The count sheet shows item name, SKU, location, and the system's expected quantity — but keep the expected quantity hidden from the counter until after they've counted. Bias toward the expected number is one of the most common counting errors. I've seen counters "round up" to match the system and introduce phantom stock that persists for months.

Step 3: Count physically

The counter goes to each location and physically counts what's there. No guessing. No rounding. No skipping items that are hard to reach.

Step 4: Reconcile vs. system

Enter the physical count into your system. The system compares the actual count against the expected quantity and flags any variance.

Step 5: Investigate variances

Any discrepancy beyond your tolerance threshold (typically ±2% by value) should be investigated before the count is finalized. Common causes: miskeys, receiving errors, items moved to wrong locations, unreported waste or damage.

This is the step most manufacturers skip when they're busy. Don't. The investigation is the point. The count itself just surfaces the number — the investigation tells you why it's wrong and how to fix it.

Step 6: Update the system

Once variance is explained or confirmed, update the system with the accurate count. Document the reason for any adjustment.

Common Mistakes to Avoid

Skipping variance investigation. The investigation is the most valuable part. If you simply accept the physical count without understanding why it differs from the system, you'll never fix the root cause — and the same error will recur.

Counting too infrequently. A items counted monthly aren't really being managed as A items. Stick to the frequency your tier requires.

Not assigning ownership. Cycle counting works best when specific people own specific areas or categories. Accountability drives consistency. "Everyone's responsible" means no one is.

Announcing counts in advance. If staff know a count is coming, they may make informal adjustments that mask real problems. Unannounced counts are more accurate.

Counting without hiding system quantities. Show counters the expected quantity and they'll count to it. Always record the physical count before comparing to system records.

What I Wish I Knew Starting Out: Shrinkage Reveals Itself

When manufacturers run their first real cycle count program, they almost always discover shrinkage they didn't know existed. Not theft necessarily — unexplained loss from damage that wasn't logged, product used informally, receiving errors that went unnoticed.

A 20-person food manufacturer I worked with found $6,000 in annual shrinkage after their first quarter of cycle counting. It was concentrated in one raw material category. They traced it to a storage area with a humidity problem that was degrading a specific ingredient. Annual physical count? Never would've surfaced it — the full year's loss happened gradually, and the end-of-year count just validated whatever was left.

That's the real value of cycle counting: it makes the invisible visible, early enough to act on it.

Key Metrics to Track

Inventory accuracy rate — the percentage of items counted that match system records within your tolerance. Target: 95%+. Calculate as: (items with no variance / total items counted) × 100.

Variance rate — the percentage of items with discrepancies. Track over time to see whether your processes are improving.

Shrinkage rate — the monetary value of inventory lost to theft, damage, or unexplained variance as a percentage of total inventory value. Cycle counting surfaces shrinkage that annual counts miss.

Review these metrics monthly and use trends to guide process improvements. A rising variance rate is a signal to investigate — not ignore.

Getting Started

Cycle counting doesn't require a complex rollout. Start with your A items. Set a weekly schedule. Assign an owner. Run it for a month and review the results. Then expand to B and C items as the process becomes routine.

One practical tip: pick a consistent time for your A item counts — same day, same time each week. Consistency builds the habit, and the habit is 80% of the battle.

For manufacturers using Nstock, cycle count scheduling, variance tracking, and ABC classification are built in. Explore the features → | See how other manufacturers use Nstock → | Get help setting up cycle counting →

— Sarah Chen