Acumatica gives you real locations inside a warehouse, with quantity tracked by location. That puts it ahead of where a lot of mid-market ERPs start, and it's a big part of why companies pick it.
The catch is that a location takes about fifteen seconds to create, so setup happens fast and early, without much argument. Then a year of transactions piles up against those records and the cheap decisions get expensive. These are the seven I slow people down on.
1. Naming
Location IDs sort as text. Name them A1, A2, A10 and A10 lands between A1 and A2, so every pick list you print for the life of the system sends someone down the aisle in the wrong order.
Use fixed width, zero padded, in the order a person walks: A01-14-3-2, with segments for aisle, bay, level and position. You can rename later, but a year of receipts and transfers point at the old names and your team has memorized them. Renaming in a live building is a weekend of relabeling and months of people saying the old name out loud. Do it before go-live. It's the best hour you'll spend on setup.
2. What counts as available
Each location has flags that decide whether its quantity is available to promise and whether you can pick or ship from it. This setting causes the most confusion later, because it decides whether on-hand and sellable agree. Quarantine, QC hold, damaged goods and returns generally shouldn't feed availability. Stock that's in the building but broken or unreleased, and that customer service can still promise, is how you ship something you shouldn't.
Decide per location type and write down why. When someone asks in six months why the warehouse total and the available total are four hundred units apart, you'll want a better answer than a shrug.
3. Pick priority
Priority drives the order Acumatica suggests pulling from. Set carelessly, pick lists zigzag. Set well, they walk a picker one way through the building and empty the pick face before the reserve. Usually that means pick faces before bulk and fast zones before slow, in the sequence of the physical route rather than the order the locations were created. If you run lots with expiry, priority interacts with allocation, so test it on a real example.
4. Default locations
Items can carry a default location per warehouse, and the warehouse carries receiving and shipping defaults. Set right, receiving is fast because the system proposes the home and the receiver confirms it. Set wrong, everyone overrides the default on every line, which is slower than having no default and teaches people to ignore the screen. Set defaults for the items that move. The long tail can go to a general receiving location and get put away on purpose.
5. Lot and serial tracking
This is the hardest one to reverse. Switching an item to lot or serial tracking once it has history and on-hand isn't a checkbox. It usually means zeroing the item out, changing the setting and bringing stock back in with tracking filled in, which is something you schedule, not something you do on a Tuesday afternoon.
Decide per item before you load anything, against a real requirement: a customer contract, a regulator, a recall exposure you can describe. Turning it on because it seems prudent adds handling time to every receipt and pick forever, and I've watched teams quietly bypass it within a month. Then you have the cost and none of the traceability.
6. Replenishment
Acumatica can move stock from bulk to the pick face on min and max by location. It's worth setting up second. Get put-away and picking stable first, then turn on replenishment once you trust the quantities. Rules built on location data that's still settling generate tasks nobody believes, and once a team learns to ignore a queue, winning that trust back takes far longer than the setup did.
7. How many locations on day one
One temptation is to model every shelf. The other is to create one location called MAIN and put the problem off. MAIN is worse. It looks like a quick go-live, but your quantities carry no position at all, so you get the overhead of a location-aware system with none of the benefit. Unwinding it later means adding locations to items that already have history.
Go live with a modest, correct scheme covering the areas that matter and add detail in month four. Pick faces, bulk racking, receiving, staging and whatever quarantine you need is usually enough to run on.
Before you configure anything
Walk the building with the person who picks in it and draw the route. Write the location names on the drawing, check that the sequence matches the walk, then check that a stranger could find A03-07-2-1 without help. Nearly every location problem I get called about started as a scheme designed at a desk by someone who hadn't made that walk.