A second warehouse usually shows up for a good reason. Orders start arriving from a region the original building can't reach in two days, freight costs climb, and someone runs the numbers and finds it's cheaper to hold stock near the customer than to ship it across the country. That's growth, not overreach. What catches teams out is how much changes the moment inventory lives in two places instead of one.
What the second location buys you
Shipping from whichever site sits closest to the customer takes a day or two off transit and real money off the freight bill. It also gives you somewhere to put stock when the first building runs out of rack, which is often the reason the conversation started in the first place. And when a storm closes one facility or an inbound container sits at anchor, you have another shelf to pull from instead of telling a customer their order is on hold.
What gets harder
Every problem below traces back to the same thing. Your inventory record now has to be right in two places at once, and nothing about that happens on its own.
Keeping both sites stocked. Attention drifts toward the busier building. The quieter site runs down slowly, nobody notices because the company-wide on-hand number still looks healthy, and the shortage announces itself as an order that can't ship.
Getting orders to the building that has the stock. Customers deal with one company and have no idea which of your sites holds the item. When an order lands on a warehouse that's out, you either transfer it, which costs time and freight, or split the shipment, which costs margin and usually irritates the customer.
Reconciling stock that's between sites. A transfer that has left one building and hasn't been received at the other belongs to neither count. If nobody owns that gap, it turns up at year end as shrink that isn't really shrink.
1. Put every location on one system of record
This is the technique the other three depend on. If each building keeps its own count in its own spreadsheet, everything downstream is guesswork.
One system with location-level detail lets you see on-hand by site rather than one blended company number, move stock between buildings as a recorded transfer with a status rather than an email, and route an order to a site that can actually fill it. It also means the answer to "do we have twelve of these anywhere" takes seconds instead of two phone calls.
The practical test is whether someone standing in the aisle with a handheld sees the same number a customer service rep sees on their screen. If those two numbers can disagree, you don't have one system of record yet.
2. Lay out each building around what it actually ships
The two warehouses probably don't ship the same mix, so they shouldn't be laid out the same way. Pull the last few months of order lines per site and put the fastest movers closest to the pack and ship area at each one. Slow movers can go up high and far back.
Label bins and shelves clearly enough that someone who has never worked that aisle can find the location from the pick list without asking. That's the difference between a new hire being productive on day two and being productive in month two.
For sites that bring in seasonal team members during a peak, post a printed map of the layout at the dock door and at the end of each aisle. It costs nothing and it saves a surprising number of walking miles.
3. Bring suppliers into the same system
Once purchasing can see supplier availability, pricing, and lead time next to the on-hand figure for each location, reorder decisions get much less dramatic. You can see that the item is short at the east warehouse, that your primary supplier is three weeks out, and that a second supplier can be there in five days, all before anyone has picked up the phone.
Lead time is the field people skip and then regret. A reorder point calculated without it will always be wrong at the site that's furthest from the supplier.
4. Count by scanning, not by walking the aisles with a clipboard
Nobody is doing a full physical count of two buildings often enough to keep the numbers honest, and the annual count that does happen tends to get pencil whipped when the clock runs out.
Scanning stock as it's received updates the record at the moment it lands, which is the only time anyone is genuinely certain what's on the pallet. From there, cycle counting a slice of locations each week keeps accuracy up without shutting anything down. Count the fast movers and the high-value items more often than the rest, since those are where an error costs you an order.
The gain compounds across sites. When receiving is scanned at both buildings, a transfer stops being a leap of faith and becomes a record you can check.
Where to start
If you're standing up a second location now, get the single system of record in place before the first pallet moves. Retrofitting it after both buildings have developed their own habits is considerably more work than doing it up front. If you're already running two sites and the numbers don't agree, start with scanned receiving at both, because everything else you'd like to fix depends on the incoming count being right.