How Shipsy automates cycle counting and keeps inventory accuracy above 99%

Shipsy’s WMS replaces the once-a-year wall-to-wall physical count with continuous, risk-weighted cycle counting. The engine decides what to count, when to count it, and who counts it — based on SKU velocity, historical variance, event triggers, and regulatory requirements. Most warehouses hit sustained inventory accuracy above 99% within a quarter of rollout.

Why we built this

Annual physical counts are theater. Everything stops for a weekend, finance signs off a one-point-in-time number, then accuracy decays over the next 12 months. Operations teams routinely discover they’re 5-10% off reality three months after the annual count. A global pharma CDMO handling multi-country clinical supply, a major Middle East 3PL and contract logistics provider, and a MENA retail conglomerate with 80+ years operating multi-brand portfolios all identified inventory accuracy decay as a top-five operational risk — driving stock-outs, mis-promises, and margin loss.

Shipsy’s continuous cycle-counting engine treats accuracy as a live KPI, not an annual audit.

How it works

ABC velocity tiering. Every SKU is classified A (fastest movers, ~10-20% of SKUs, 60-70% of volume), B (mid-velocity), or C (slow), updated continuously as velocity shifts. Count frequency scales with tier: A-SKUs counted weekly or bi-weekly, B-SKUs monthly, C-SKUs quarterly.

Risk-weighted count scheduling. Tier isn’t the only signal. The engine layers additional risk factors: - Historical variance — SKUs with a track record of shrinkage or mis-counts get higher count frequency regardless of velocity. - Recent events — SKUs involved in recent returns, damages, or receiving disputes trigger targeted counts. - Location risk — bin locations with cross-tenant exposure (3PL multi-client) or access points get higher count coverage. - Value — high-unit-value SKUs get tighter count cadence than low-value even at the same velocity tier.

Zero-count trigger. When a picker scans a bin and finds it empty while the system expected inventory, the engine fires an immediate count task for that bin. Zero-counts catch variance in near-real-time rather than weeks later.

Negative-count trigger. When a pick would drive a bin negative (picker tries to remove more than system says exists), the engine flags and triggers both a count task and a receiving-audit trail to identify whether the source of the variance was receiving, putaway, or shrinkage.

Count task dispatch. Count tasks are injected into the daily task queue alongside picks, puts, and replenishment. Counters scan bins, the engine compares count to expected, and variances are either auto-adjusted (small variances within tolerance) or escalated to a supervisor for root-cause review. Large variances trigger a second blind count by a different counter before any adjustment is applied.

Root-cause classification. Every variance is classified: receiving mis-count, putaway mis-location, pick mis-count, shrinkage, damage, or system error. Pattern detection surfaces systemic issues — e.g., “three variances this week all trace back to receiving dock 2 on Tuesday shift” — so operations can act on root cause, not symptoms.

Compliance-grade audit trails. For regulated warehouses (pharma GDP, 21 CFR, high-value goods), every count event carries counter ID, timestamp, photo evidence, and adjustment approval chain. A global biotech scaling rare-disease therapeutics across 30+ countries uses Shipsy’s cycle-counting trail as the audit record satisfying regulatory inspection requirements. See pharma 21 CFR / GDP compliant warehousing.

Multi-tenant segregation. In 3PL multi-tenant warehouses, cycle counts are scheduled and reported per tenant — each client sees accuracy metrics for their own inventory only, with tenant-level SLAs tracked separately. See 3PL multi-tenant warehouse operations.

Exception escalation. When cumulative variance on a SKU or bin crosses a threshold, the engine escalates to a full-bin audit — all adjacent bins counted together to catch misplaced inventory. This is far more effective than counting the same bin twice and finding the same wrong number.

Continuous accuracy reporting. Inventory accuracy is a live KPI visible at SKU, bin, zone, and warehouse levels. Finance, operations, and client-facing teams see the same number, updated daily. This eliminates the “operations says 98%, client says 92%” gap that plagues multi-tenant warehouses.

Early results

Warehouses moving from annual counts to continuous risk-weighted cycle counting typically achieve: sustained inventory accuracy above 99% (vs. industry-typical 94-97%), zero annual-count shutdown downtime, and 30-50% reduction in emergency count events because variance is caught and resolved inside normal operations. A major Middle East 3PL and contract logistics provider runs this across multiple client tenants and uses tenant-level accuracy as a differentiator in new-tenant contract wins.

For pharma and high-value operations, the compliance benefit is material — audit inspections move from week-long preparations to pulling existing cycle-count evidence.

What’s next

Next release adds computer-vision-assisted counting for shelf-level verification — CV cameras on forklift masts or ceiling mounts detect bin discrepancies during normal travel, feeding findings into the count-task queue automatically. This reduces manual count labor while increasing coverage density.