Most inventory errors are born at the dock

When accuracy is bad, the investigation almost always starts at picking. That is where the customer feels it, so that is where people look.

I have traced a lot of variances back to their origin, and the pattern is consistent. Picking errors are real and they are usually singles: one unit, one order, one annoyed customer. Receiving errors come in case quantities and they sit undiscovered for weeks.

An item mis-received at 24 instead of 20 is wrong by four units for however long it takes to count that location. Every report built on that number in the meantime is wrong too.

The dock is where accuracy is made.

The count you skipped is the variance you will find in November

Almost every warehouse under pressure has a version of the same shortcut. The truck is late, the driver is waiting, so the packing slip gets signed and the pallets get counted later. Later does not always happen.

I am sympathetic to why this exists. What I push back on is the idea that it is free. It is a loan, and the interest is paid in cycle count variances that take an hour each to investigate three months on.

Two ways out that do not require software.

Separate receipt from putaway. Signing for the freight and counting it do not need to be the same event. Take the freight into a staging location that the system knows about, release the driver, and count against the packing slip within the shift. The stock is on the books in a location you can find, and nobody is standing over you.

Count what varies. Not every line needs a piece count. A sealed case from a vendor with a clean history can be counted as cases. A vendor with a history of short shipments gets full counts on every line until that changes. Build the list from your own variance history, review it a couple of times a year, and tell the vendors on it that they are on it. That last part fixes more than the counting does.

Blind receiving, at least for the problem vendors

If the receiver can see the expected quantity, the expected quantity is what gets entered. This is not dishonesty. It is what happens to anyone who counts to 19, sees a 20 on the screen, and assumes they lost count.

Blind receiving means the receiver enters what they counted, and the system compares. It is slightly slower and it catches short shipments you are currently paying for.

You do not have to do it everywhere. Run it on the vendors and the items where your variance history is worst.

Putaway is where the location gets lost

A correct count that lands in an unrecorded location is still an error, and it is a nastier kind because the total is right and the stock is not findable.

The failure looks like this: the pick face is full, so the pallet goes into the first open reserve slot, and the intention is to record it after lunch. The system says the stock is in the building. Picking goes to the pick face, finds it short, and reports a stockout on inventory you own.

The habits that prevent it are dull.

• One transaction moves stock, and it records both where it left and where it landed. No moves that only record a quantity.
• Nothing gets set down in a place without a label. If there is floor space where pallets regularly sit, give it a location code and let it be real.
• The person who moves it records it, at the time they move it. Not a runner with a clipboard at the end of the shift.

If you have scanners, this is close to free. If you do not, it is a paper move ticket that goes in a box by the office and gets entered daily, which is not elegant but does work.

Directed putaway is worth less than people think, at first

Vendors demo directed putaway well. The system decides the best slot, the forklift driver goes where told, everything is optimal.

It is genuinely useful at scale. Early on, it is only as good as your cube data and your slotting rules, and if either is rough it will send drivers to bins that will not hold the pallet. Two or three of those and the drivers stop trusting it, which is very hard to undo.

I would rather see a company go live with simple fixed home locations that are correct than dynamic putaway that is clever and wrong. Turn the smarter rules on in month four, once the cube data has been proven by actual use.

Returns are receiving too

Returns come back through a different door, get handled by a different person, and follow a process that in a lot of warehouses is mostly improvised.

The common failure is stock going back on the shelf before it is received, which means the shelf has more than the system does. The opposite failure, a credit issued for stock that never physically arrived, is rarer and more expensive.

Give returns their own location, receive them properly, and make the disposition decision (restock, quarantine, scrap) an explicit step with a transaction behind it. Returns volume is small enough in most SMB warehouses that this takes very little time and prevents a category of variance that is otherwise almost impossible to trace.

What to measure

If you want one dock metric, use receiving variance rate: lines where the received quantity differed from the count, divided by lines received. Track it by vendor and by receiver.

By vendor, it tells you who to have a conversation with, and gives you the data to have it with.

By receiver, it tells you whether you have a training issue or a process issue. If everyone's rate is similar, the process is the problem. If one person's is triple everyone else's, that is twenty minutes of coaching.

Add a second one if you can: hours from freight arrival to putaway complete. Long dwell time in staging is where stock becomes invisible, and it tends to creep upward quietly until someone starts watching it.

Neither number requires software you do not already have. Both will point you at the actual source of the accuracy problem you have been chasing at the pick face.