A cycle count that stays right between counts.
A cycle count finds the error, not why it keeps happening. We encode your count and variance rules into the systems your stores already run.
Official services partner of the platforms defining AI
NVIDIA
Anthropic
The counts come back wrong. The reason is never written down.
From independent hardware and building-supply co-ops we have worked inside: 20–25-store operations on Epicor Eagle-class POS/ERP, where a month’s counts land on stores all at once.
Error rates that never move
“30–40% count error rates.” Managers hand-key variances from paper count sheets into the PIP file — the same items come back wrong next month.
A month of counts, dumped at once
“13-page co-op PDFs processed with red/green dots.” A month’s counts land in one drop; by the time they’re keyed, whoever could explain a variance has moved on.
Dead stock nobody will buy
“Hundreds of thousands of dollars of product no one will buy.” Unpicked-up special orders sit on the shelf and in the system, inflating an already-wrong on-hand number.
75% accuracy against a 95% bar
“75% inventory accuracy vs 95% best-in-class.” At that level the on-hand number is a guess, so every purchase, transfer, and reorder built on it inherits the error.
The vendor-credit window closes
Shorts and damages are real, but the credit window is 30 days. Blind-received items go missing past the freight-claim window, and the loss moves straight to the P&L.
The error compounds between counts.
Measured inside independent hardware and building-supply operations. Your numbers will differ — the one-store variance audit puts figures on yours before anything gets built.
From anonymized engagements — independent hardware and building-supply co-ops
We encode your exceptions.
No new system for your stores to learn. The count rules your best inventory manager carries in their head become the rules the system enforces.
- 01
Find where the count breaks
We audit one store for one week, tracing every variance to its cause: tolerances by ABC class, unit-of-measure traps, location logic, receiving cutoffs. You get the map first.
- 02
Encode the rules
Tolerances by ABC class, case-vs-each conversions, and exception queues become rules in the systems you already run. Clean counts flow; violations route to their owner.
- 03
Count by exception
Humans stop counting everything. The rules resolve what they can; your people see only the variances the rules can’t explain — a short queue, not a full count.
A counting service tells you what you have. It can’t fix why it’s wrong.
Rent-a-counter firms send a crew, hand you a number, and leave. Thirty days later the number is wrong again, because nothing about how items get received, keyed, or moved has changed.
New software won’t fix it either. We encode your count and variance rules into the systems your stores already run, and our AI reconciles every count against them — so accuracy holds between counts instead of resetting when a crew walks out.
Our counts were running 75% accurate and nobody could say why. Now the system flags the variance and my managers only touch the exceptions.
Led by Peter Enestrom.
Asked by owners and ops managers.
The straight answers, before you book anything.
A cycle count is a recurring count of a rotating slice of inventory — a bin, an aisle, an ABC class — instead of one full physical inventory. The goal is a perpetual check on the system’s on-hand number, so errors surface in days instead of at year-end.
A full physical inventory counts everything at once, usually once a year, and freezes the store to do it. A cycle count samples on a schedule — fast-moving or high-value items more often — so accuracy is maintained continuously instead of rebuilt annually.
Best-in-class is 95% or better, item by item, location by location. Many multi-store operators run closer to 75%, which means one record in four is wrong and every purchase built on those records inherits the error.
Almost never theft first. The causes are process: units of measure keyed inconsistently, items blind-received past the freight-claim window, receiving cutoffs that differ by store, variances keyed from paper. A variance audit traces one week of mismatches to the exact rule that was never written down.
By class, not by calendar. Fast-moving A items weekly or monthly, C items once or twice a year. When the count rules are encoded, the schedule becomes exception-only: the system resolves what it can, and people count only what the rules can’t explain.
Start with one workflow.
Tell us where your team loses hours. We will come back with a straight answer on whether AI can help, what it would take, and what it would pay.