Running multiple sites and companies without a second source of truth

How to tell a physical location apart from a legal entity, move stock between sites so the lot stays traceable, keep each company's books separate while operators still see across them, decide what data is shared versus walled off, and forecast and buy across more than one location.

After this lesson you can draw the map of your operation as it really is — which physical places hold stock, which legal entities own it, and what moves between them — and set the handful of rules that keep it coherent as it grows: a transfer that carries its lot from one site to the next, books that stay separate per company while operators still see across them, and a deliberate line between the data you share everywhere and the data you wall off. The aim through all of it is one thing: as you add a location or a company, you do not add a second version of the truth.

A location and a company are two different questions

Growth adds complexity along two axes that people tend to blur, and untangling them first keeps everything after it from knotting up.

A is about where. A is about who owns it. The two vary independently. You can run two plants inside one company. You can have one plant shared by two companies. You can have the full grid of several locations across several entities.

Keeping the two apart matters because the rules differ. Moving a pallet between two of your own locations is a warehouse move — the same owner, both ends. Moving that pallet between two of your companies is a change of ownership — one entity's books lose it, another's gain it, at a price. Same pallet, very different record. Get the axis wrong and you either lose track of where stock is or you quietly mix up whose stock it was.

Multi-location inventory: where the stock actually is

With one location, on-hand is a single number. Add a second and every on-hand splits: 40 cases here, 12 cases there. The trap is treating it as one pile — the company-wide total reads healthy while one site is out and the other is overstocked, and nobody notices until a run stops for a material that was "in stock" at the wrong address. So inventory has to be counted and held per location, not just per company.

Moving stock from one location to another is a . The discipline that keeps the record honest is to treat it as exactly those two events. Stock leaves location A and is not yet at location B; it is in transit, belonging fully to neither shelf. Only when it is received at B does B's on-hand rise. Skip the receiving half and the record says stock is in two places at once, or in none — and the on-hand drifts from the shelf the same way a missed count does. For one concrete way to record that move as the two linked events it is — a shipment out of one location and a receipt into the other — the docs on moving stock between locations walk it.

Keep the lot on the transfer, or the trail breaks at the door

A transfer is more than a warehouse chore, because the lot travels with the stock. If the transfer record does not carry the lot, the genealogy snaps the instant the stock crosses to another site — B receives a pallet with no history behind it.

So a good transfer captures the lot code, quantity, and expiry as the stock ships from A, and confirms them as it is received at B. That is the same one-step-forward, one-step-back shape you already keep inside a single location, now stretched across the gap between two. Do it and a trace still runs end to end: this finished lot moved to site B on this date, and shipped from there to this customer. The rotation rule has to survive the move too — the expiry rides along so the receiving site keeps shipping soonest-to-expire first, not oldest-arrived-first, because from B's point of view everything "arrived" the day the transfer landed.

For covered foods, the transfer records are also a rule

When food on FDA's Food Traceability List changes hands — you receive from a supplier, ship to a customer, or transfer between two of your legal entities — the Food Traceability Rule (FSMA 204) attaches specific shipping and receiving records to that move, and each entity keeps its own half. Whether the rule reaches a given food and a given move depends on the food and the transfer; the FSMA-204 lesson covers what to record, and FDA's Food Traceability List and rule page carry the current covered-food list and compliance date (extended to July 20, 2028; verified July 2026). Keep the transfer discipline whether or not the rule binds you — the trace is the point, the rule is the floor.

Multi-company: separate books, shared visibility

The second axis bites when the growth is a new legal entity, not just a new address. Companies get split for real reasons — a separate entity for a new brand, a co-packing arm, a retail arm, a partner with different owners, liability kept apart, taxes filed separately. Each entity keeps its own books: its own purchase orders, its own invoices, its own financials, its own cost of goods. That separation is not an inconvenience to work around; it is the whole reason the entities exist.

And yet, standing on the floor, you still want to see across all of them at once — where inventory sits, what is on order, what a batch costs — without logging into two disconnected systems and reconciling them by hand at month-end. So multi-company pulls in two directions at the same time: the books stay walled per entity, while the visibility stays shared across them.

The move between two of your companies is the one to name carefully. It is not a warehouse transfer — it is an . Company A sells or transfers the stock to Company B at a real price, and both sets of books have to reflect it. Record it as a plain warehouse move and you have understated one company and overstated the other — the books stop being clean, which was the reason to separate the entities in the first place.

Shared or walled: draw the line on purpose

The central decision in running more than one site or company is, for every kind of data, whether it is shared across the whole group or walled inside one entity. Two failure modes frame the choice.

Everything walled. Each site and company is its own island. You re-key the same product, the same recipe, the same supplier into each one; nobody can see across; and the single source of truth the whole setup was meant to protect is gone, replaced by several partial truths that drift apart. This is where a team ends up as the integration layer, copying numbers between systems that should have been one.
Everything shared. The boundaries that legally and financially must stay separate leak. One company's financials, pricing, or customer list bleeds into another's, and the clean books you split the entities to keep are muddied again.

The line most operations settle on: share the operational backbone — the product catalog, recipes, units, the approved-supplier list, and read-only visibility of inventory and lot genealogy across sites — and wall the financial and ownership layer — each entity's books, its pricing, and per-entity roles and access. Share what everyone needs to run the floor; wall what the law and the accountant need kept apart.

How cleanly a system lets you draw that line depends on how it was built. Some systems assume a single company; to run a second you stand up a second copy and wall it off entirely, then reconcile the two by hand — visibility lost to keep the books apart. Others are built , so the line between shared and walled is one the system already understands rather than one you rebuild by hand. Bettr Manager, the operations platform this site is part of, is built that way; it is one example of the multi-company-by-design approach, not a verdict on which system fits your operation. The decision that stays yours either way is the same: list your data types and mark each one shared or walled before you shop, so you can tell whether a given system can actually hold the line you need.

Forecasting and purchasing across sites

The last thing multiplication changes is how you buy, and it changes in two opposite-looking ways at once.

Reorder points run per location. Each site has its own usage rate, its own lead time, and its own safety stock, because a material burns down faster at the busy plant and the truck takes longer to reach the far one. A single company-wide reorder point hides both — it fires late for one site and early for the other. So the reorder-point math you set for one location gets kept per location, one set of numbers each.

But demand can be forecast and bought across all of them. Add up what every site will use and you can put the total on one purchase order to earn the volume price, then transfer to each site as it needs it — buy centrally, hold locally. Take three sites each using around ten cases a week of the same oil: bought as a single order of thirty they price far better than three separate orders of ten, and the savings land on every recipe that oil touches. (Those figures are illustrative — the point is the shape, not the numbers.) The approved-supplier list is one of the things you share rather than wall: a supplier vetted for one site is vetted for all, and three separate lists would only drift apart.

None of that consolidation works unless you can actually see across the sites. If the visibility is walled off, every site is back to buying small and paying retail, guessing at its own demand alone — which is exactly the case for keeping inventory visibility on the shared side of the line.

Where this leads

Everything here assumes the sites and companies are yours. You control both ends of every transfer, you set the rules on both sides, and the whole map lives in systems you own. The moment production crosses into a partner you do not own — a co-packer making your product, or your plant making someone else's — the same transfer-and-record discipline has to hold across an ownership line you only half control. Who keeps which lot record, who can release or hold product, and how lot data passes with every shipment in both directions become things you pin down in a contract, not just a settings screen. That is the next lesson: co-manufacturing and release chains.

Do this in your operation

Draw the real map. List every physical location that holds stock and every legal entity that owns it, and mark which entity owns the stock at each location. Note every place the two lines do not line up — a location shared by two companies, or one company spread across two sites.
For each pair of places stock moves between, write down whether the move is a within-company transfer (a warehouse move) or an intercompany transfer (a change of ownership at a price). They are recorded differently — know which is which before you record one.
Define how a transfer is recorded so the lot survives it: the lot code, quantity, and expiry captured as stock ships from one place and confirmed as it is received at the other. Then trace one lot end to end across a transfer to prove the trail does not break at the door.
List your data types — product catalog, recipes, units, approved suppliers, inventory visibility, lot genealogy, financials, pricing, customer lists, user access — and mark each one shared across the group or walled inside one entity. Decide it on purpose, before a system decides it for you.
If any of your food is on FDA's Food Traceability List, confirm at the source what shipping and receiving records FSMA 204 asks for on the moves that cross a legal-entity line, and which entity keeps which half.

Mark this lesson complete