Part 5 of our series on making stock takes quicker and more accurate
Your system says there should be forty-seven units in the bin.
You count. One, two, three... you get to forty-four.
But you know there should be forty-seven. So you count again.
This time you find forty-seven.
Did you miscount the first time? Or did you unconsciously adjust the second time to match what you expected?
You'll never know.
That's the problem with counting when you already know the answer.
Your counter isn't lying.
They're not being careless.
They're being human.
When you see the number forty-seven on the screen before you count, your brain starts looking for forty-seven. It groups items into patterns that add up to forty-seven. It dismisses counts that don't match as obvious mistakes.
The psychologists call this confirmation bias.
Warehouse managers call it Thursday.
Because it happens every single time someone counts while looking at the expected quantity.
The solution isn't to try harder. The solution is to stop looking.
Blind counting sounds simple.
Don't show the counter the system quantity.
But most warehouse systems aren't built for this.
The expected quantity is right there on the screen. The count field is right below it. You can't unsee what you've already seen.
Real blindness requires deliberate design.
Workflow: A bearing distributor reprograms their RF scanners to support blind counting. When a counter scans a location, the screen shows only the item number and description. No quantity. They count and enter the number. The scanner stores it but doesn't display anything. Then they hand the scanner to a second person who pulls up the system quantity and compares. Only at that moment does anyone see both numbers together.
Example: A chemical manufacturer without fancy scanners uses a paper system. Counter A gets a list with item numbers and locations, but no quantities. They count and write their number on the sheet. Counter B has a separate clipboard with the system quantities. After Counter A finishes each item, Counter B looks at both numbers. If they match, they initial it. If they don't match, they circle it for recount. The sheets never combine until the end of the shift.
You can't be blind if you saw the answer first.
Two people counting sounds like double the work.
It's not.
Because they're doing different jobs.
Person one is counting. Physical work. Eyes on the inventory. Touching boxes. Verifying labels.
Person two is verifying and recording. Mental work. Comparing numbers. Flagging discrepancies. Triggering recounts.
When the roles are clear, the work flows.
Workflow: An electronics distributor creates a formal role definition. The "counter" never touches the scanner or sees the screen. Their only job is to physically count and call out the number. The "verifier" holds the scanner, sees the system quantity, and records the count. If there's a discrepancy over their threshold (five percent), the verifier says "recount please" without revealing what the system expects. They recount together, still blind, until they agree on the physical quantity.
Example: A food ingredients warehouse pairs an experienced person with a newer employee. The experienced person is always the verifier because they understand when variances make sense versus when they signal a bigger problem. The newer employee does the physical counting, which helps them learn the inventory and warehouse layout. After three months, they swap roles. Everyone learns both sides.
Division of labor isn't just about efficiency. It's about focus.
Your business is unique — your software should be too. Let's talk about a system built around how you actually work.
Get a free quoteNot everyone works well together.
Some people rush each other. Some people slow each other down. Some people defer too much. Some people dominate.
The wrong pairing turns blind counting into a fight or a rubber stamp.
The right pairing turns it into a system that catches errors.
Workflow: A machine shop experiments with different team pairings over a month. They track not just accuracy but also speed and employee feedback. They discover that pairing their fastest counter with their most detail oriented verifier creates the best results. The fast counter keeps the pace up. The detail oriented verifier catches the subtle errors. They also learn that pairing two detail oriented people creates gridlock, while pairing two fast people creates sloppiness.
Example: A pharmaceutical distributor has a senior warehouse worker who's been there fifteen years. He knows every product and can spot a mislabeled box from across the aisle. They pair him with younger workers who are learning the inventory. He verifies, they count. His expertise flags the weird variances immediately. "That number can't be right, we never stock that item in quantities over twenty." They recount and find the error. The younger workers learn what normal looks like.
Chemistry matters more than credentials.
The counter says thirty-two.
The system says thirty-five.
Now what?
If the verifier just says "wrong, count again," the counter gets defensive.
If the verifier says "the system shows thirty-five," you've destroyed the blindness.
You need a protocol that allows disagreement without revealing the answer or creating conflict.
Workflow: A fastener manufacturer creates a simple script. When there's a variance, the verifier says "outside tolerance, recount requested." Not "you're wrong." Not "the system disagrees." Just a neutral request based on a threshold. The counter recounts without knowing what number they're trying to hit. If the second count matches their first count, they say "confirmed thirty-two." Now the verifier knows the physical count is probably accurate and the system is probably wrong.
Example: A building materials supplier uses colored flags. Green flag means the count matches the system, keep moving. Yellow flag means minor variance within tolerance, the verifier will investigate later but they move on. Red flag means significant variance, recount now. The counter never knows which color corresponds to their number. They just know red means count again, carefully. This removes the emotional charge from being "wrong."
Disagreement isn't failure. It's the system working.
A three unit variance on an item you stock in the thousands doesn't matter.
A one unit variance on an item worth five thousand dollars matters a lot.
If you recount everything that's off by even one unit, you'll spend all day recounting.
If you accept all variances under ten percent, you'll miss important errors.
Smart tolerances let you move fast on what doesn't matter and slow down on what does.
Workflow: An industrial supply company sets three different tolerance levels tied to their ABC classification. A items: recount any variance. B items: recount if variance exceeds two percent or five units, whichever is less. C items: recount if variance exceeds ten percent or twenty units, whichever is less. The verifier sees the item class on their screen and knows which rule to apply. This keeps them moving through C items quickly while being rigorous on A items.
Example: A steel service center uses dollar based tolerances instead of percentage based. Any variance worth more than one hundred dollars triggers an immediate recount regardless of unit quantity. A variance of two tons of rebar worth three hundred dollars gets recounted. A variance of fifty pounds of screws worth twelve dollars gets accepted. They're managing the financial impact, not just the numbers.
Perfect counts everywhere is the enemy of accurate counts where it matters.
Here's what usually happens when someone counts wrong.
They count and get thirty-two.
The system says thirty-five.
They count again and get thirty-five.
They enter thirty-five.
Everyone's happy.
Except now you have no idea if the system was right or if the counter just stopped trying.
The first count is the most honest count. Preserve it.
Workflow: A automotive parts distributor requires the verifier to record both the first count and the recount result. Their system logs: Item 4471, System quantity 47, First count 44, Recount result 47, Final variance 0. Later, they analyze this data. If someone consistently matches the system on the second count but not the first, that's a red flag. Either they're a sloppy counter who gets better under pressure, or they're unconsciously adjusting to match.
Example: A packaging manufacturer discovers a pattern in their recount data. One counter's first count is wrong sixty percent of the time, but their recount matches the system ninety-five percent of the time. Investigation reveals he's peeking at the verifier's screen during the recount. They move the verifier five feet away during counts. His recount accuracy drops to seventy percent, which is actually more honest. They provide additional training on counting techniques.
The pattern in the data tells you more than any single count.
Your team needs to understand that confirmation bias isn't a character flaw.
It's how human brains work.
When you tell someone "don't let the system number influence your count," they nod and say "of course not."
But they don't believe they're actually doing it.
Because bias is invisible to the person experiencing it.
Workflow: A electrical supply warehouse runs a quarterly training exercise. They set up ten bins with known quantities but enter wrong quantities in a test system. They have each team member count the bins while looking at the wrong system numbers. Then they reveal the actual quantities and show everyone how many people's counts drifted toward the system number even though it was wrong. People are shocked. "I really thought I counted fifty-two, but there's only forty-eight here." Now they understand viscerally why blind counting matters.
Example: A plumbing distributor shares real examples from their own data during team meetings. "Last month we counted item 8844. System said ninety units. First counter got eighty-seven. We did a blind recount with a different person. They got eighty-seven. We recounted a third time, still blind. Eighty-seven. Then we checked the system history. Someone had done a paper adjustment two weeks ago and fat-fingered ninety instead of eighty-seven. If we'd been looking at the system number, we probably would have convinced ourselves we miscounted."
Understanding the enemy makes you better at fighting it.
Two people counting sounds like it takes twice as long.
It doesn't.
Because blind two-person counts eliminate most recounts.
When one person counts while looking at the system number, they count, second guess themselves, count again, still aren't sure, count a third time, finally accept it, and move on.
When two people count blind, they count once, verify once, flag real discrepancies immediately, and move on.
The total time is often faster.
Workflow: A food distributor times their counts before and after implementing blind two-person teams. Before: average fifteen minutes per twenty item count with one person, including the hesitation, recounts, and second guessing. After: average twelve minutes per twenty item count with two people, because they count each item once and only recount the flagged variances. The addition of the second person is offset by the elimination of unnecessary recounts and the confidence to move quickly.
Example: A hydraulics manufacturer discovers an unexpected benefit. With two-person teams, they can count during normal operations without stopping workflow. One person counts while the other handles radio calls, answers questions, and manages interruptions. Before, a solo counter would get interrupted, lose count, and have to start over. Now the counter focuses while the verifier manages the chaos. Their counts during business hours are just as accurate as their after-hours counts, which saves overtime.
Efficiency isn't always about fewer people. Sometimes it's about smarter roles.
You don't want to know what your system says is in the bin.
You want to know what's actually in the bin.
Those are different questions.
When your counter sees the system number first, they're answering the first question. "Can I confirm what the system expects?"
When your counter doesn't see the system number, they're answering the second question. "What do I actually see?"
The second question is harder.
It requires them to trust their own eyes instead of trusting the computer.
It requires them to be okay with creating a variance instead of resolving one.
It requires them to understand that finding a discrepancy isn't a failure. It's the whole point.
Because your system is wrong sometimes.
The only way to know when is to count without looking at what it says.
Your forty-seven units might actually be forty-four.
Or they might be fifty.
You won't know until someone counts them without knowing what number they're supposed to find.
That's not inefficiency.
That's integrity.
And integrity is what makes your inventory number worth believing.
Your business is unique, but your software is off the shelf? Ditch the workarounds and let's build your ERP systems to fit your teams.