Third Party Fulfillment Companies: Pick the Right Network

Third party fulfillment companies split into three operating decisions

Search “third party fulfillment companies” and you get a list sorted by name, not by fit. That is the wrong comparison to start from. Every third party fulfillment company sits somewhere on three separate axes: how many warehouses it runs and where they sit, how specialized its operation is for your product type, and whether it owns the building and the people doing the work or subcontracts to someone who does. Two providers can look identical on a sales call and behave completely differently once your order volume climbs past a few hundred a day, because the difference lives in the operating model, not the pitch deck.

For a brand doing $3M to $30M a year on Shopify Plus or a comparable subscription platform, this is not an academic distinction. It is the difference between a provider that scales with your SKU count and ship-to spread, and one that quietly caps your delivery promise the moment you cross a volume threshold nobody told you about. A returns clerk at a single-node warehouse in Ohio can hit her afternoon truck cutoff every day for a year, then miss it the week you add a West Coast marketing push and the outbound queue backs up past 6pm. Nothing about the contract changed. The order pattern did, and the operating model could not absorb it.

This article is not about which named 3PL provider to pick. It walks through the decision itself: how to read a fulfillment company’s operating model before you sign, which model suits which brand, and the setting most teams get wrong once they are live. If you are shipping fewer than a few hundred orders a month from one region with a handful of simple SKUs, most of this will not change your decision yet — read it anyway, because the thresholds where it starts to matter arrive faster than most founders expect.

What you need to know before you compare vendors

Before any vendor call, pull five numbers you likely already have. Your SKU count and which of them are simple single-unit picks versus kitted or bundled orders. Your ship-to distribution by state or region, because a network built for a Northeast-heavy customer base is the wrong shape for a brand selling evenly coast to coast. Your peak-to-average order multiplier, since a warehouse that comfortably handles your average day can still choke on your November peak if its labor model does not flex. Your return rate and whether returns need inspection, restocking or disposal decisions. And your channel mix — pure DTC, or DTC plus wholesale, plus marketplace — because each channel has different labelling, cutoff and compliance rules a warehouse either supports natively or bolts on badly.

None of these numbers need to be exact. A rough SKU count and a rough regional split are enough to have a real conversation instead of a generic one. What matters is that you walk into the first call already knowing your own shape, so you can tell whether the provider is describing its network or describing a generic pitch that fits every prospect on its call sheet that week.

Step 1: Match network shape to your ship-to spread

A single-node provider runs one warehouse. A distributed provider runs several, and routes each order to whichever node is closest to the customer, or splits it across nodes if one does not hold full stock. The mechanism that makes this matter is carrier zone pricing: ground carriers price parcels by the distance-based zone between the origin and destination ZIP code, a structure both FedEx and UPS publish in their zone charts. A single warehouse in the Midwest starts most East Coast and Gulf Coast shipments in a low zone and most West Coast shipments in a high one, so transit time and cost both stretch for roughly half the country. A three- or four-node network puts most customers inside a low zone of at least one node, which shortens transit and lowers the per-shipment carrier charge.

Volume is where this model starts to break. A single node scales fine when your order count is modest and your delivery promise is loose — three to five business days is realistic almost anywhere from almost any one warehouse. It stops scaling once you make a faster promise, or once your order volume is large enough that the extra transit day across half your customer base shows up in support tickets and repeat-purchase rate. A distributed network fixes the transit problem but introduces a new one: inventory has to exist in the right place before an order arrives, or the system falls back to shipping from a farther node anyway, or worse, splitting the order across two nodes and doubling your outbound shipping charge on that order. A brand with a handful of SKUs and steady demand can forecast well enough to keep nodes stocked. A brand with a long tail of low-velocity SKUs and a lot of new-product launches will fight stock imbalance constantly, and that fight is invisible until you pull a split-shipment report.

Peak season staffing follows the same pattern. A single-node warehouse has to flex its entire seasonal labor pool inside one building, which means recruiting, training and scheduling temporary staff against one site’s local labor market alone. A distributed network spreads that recruiting problem across several regional labor markets, which can be easier to staff in aggregate, but only if the provider coordinates hiring and training consistently across nodes. Ask how training is standardised between locations, because a customer’s order should not ship to a different quality standard depending on which node happens to fill it during your busiest week.

Step 2: Match specialization to your SKU complexity

A generalist 3PL runs one type of operation across many kinds of goods: pick, pack, ship, standard racking, ambient temperature. It works well for SKUs that do not carry a regulatory or handling requirement — apparel without batteries or aerosols, most home goods, most beauty products without flammable ingredients, books, accessories. A vertical specialist builds its warehouse, staff training and compliance program around one category: food and beverage with temperature control and lot or expiry tracking, beauty with hazmat-classified aerosols and flammables, apparel with size and colour matrix kitting, supplements with lot recall traceability.

The operational tell is what happens during a recall or a quality hold. A specialist food or supplement 3PL runs FEFO — first-expired, first-out — picking by default, and can quarantine a lot number across every node within the hour a recall notice lands. A generalist warehouse asked to do the same thing is improvising: pulling stock manually by lot, cross-checking against a spreadsheet, and hoping nothing shipped from that lot in the gap. This is not a hypothetical for a $3M-$30M consumables brand. It is the scenario a specialist contract exists to prevent, and the reason a generalist’s lower headline rate is not actually cheaper once you price in what one mis-shipped recalled lot costs in refunds, support time and, for regulated categories, reporting obligations.

The reverse mistake also happens. A brand with simple, stable SKUs sometimes signs with a vertical specialist because the sales conversation sounded more sophisticated, and ends up paying for hazmat-certified staff and temperature-controlled racking it never uses. If your SKUs do not carry a real compliance requirement, a generalist operator is very likely the better fit, and the money saved is real, not theoretical.

Step 3: Match ownership model to your risk tolerance

A broker sells the relationship and the technology layer, then subcontracts the physical work to a warehouse partner it does not own or staff. An operator owns the building, employs the people picking your orders, and runs its own warehouse management system end to end. Both can look the same in a demo, because the demo usually shows the broker’s software, not the partner warehouse behind it.

The distinction matters most when something breaks. With an operator, there is one company to call, and that company controls every lever needed to fix the problem: relabelling a mis-picked order, reprioritizing a late shipment, adjusting a cutoff for a specific SKU. With a broker, your issue has to be relayed to the operating partner, and that partner’s priorities are set by its full client roster, not by you specifically. A broker’s contract can also let it move your inventory to a different partner warehouse with limited notice, silently changing your node locations and ship-to zones without your operating model actually changing on paper.

Brokers are not automatically the wrong choice. A broker’s negotiated wholesale rate across a large partner network can beat what a smaller regional operator charges directly, and for a brand with simple, low-risk SKUs and a loose delivery promise, that price gap can matter more than the accountability gap. What you are buying either way is not just space and labor — it is who answers the phone, and what authority they actually have, at 4pm on the day a truck misses its cutoff.

Ownership model also shapes how hard it is to leave. An operator’s contract typically ties you to that one company’s notice period and inventory-transfer process, which is straightforward to plan around because there is one party to negotiate with. A broker’s contract can add a second layer: even after your notice period with the broker ends, the broker’s own agreement with its warehouse partner may set the real timeline for releasing your inventory, and that timeline is not always visible in the contract you signed. Ask directly what happens to your inventory, and how fast, if you terminate, and ask whether that answer depends on an agreement between the broker and its partner that you have never seen.

Step 4: Set the settings that decide whether this works

Settings are the step most teams skip entirely, because they live inside the WMS configuration and the contract’s operational exhibit, not the sales deck. Ask for these settings by name before you sign, and confirm the actual value, not a description of the feature.

Cutoff time. The local warehouse time by which an order must be received to ship same day. A 2pm cutoff at an East Coast node effectively means an 11am cutoff for West Coast customers checking a countdown timer on your site, because the promise has to account for the node’s own clock, not the customer’s.

Safety stock threshold. The minimum unit count per SKU per node that triggers a replenishment alert. Set too low, a fast-moving SKU stocks out mid-week and every order for it either backorders or splits to a farther node. Set too high, you are carrying and paying storage on inventory you do not need at that node.

SKU velocity tier and slotting rule. Warehouses classify SKUs A, B or C by how often they ship, and slot the A-tier SKUs closest to the pack station. A provider that cannot describe its own velocity-tiering rule is very likely running a flat slotting layout, which slows down picking on your fastest-moving products specifically.

Multi-node inventory allocation rule. This is the setting most teams never touch, and it is the one most likely to quietly damage margin. The common default is “ship from nearest node with any available stock,” which sounds reasonable until a top SKU is unevenly stocked across nodes — say, well-stocked at the East Coast node and low at the West Coast node. An order containing that SKU and another one only in stock at the West Coast node gets split into two packages from two nodes, and you pay two outbound shipping charges for one customer order, without an alert, until someone reads a split-shipment report that most teams never ask for. Ask for the split-shipment rate as a standing metric, not a one-time report, and ask whether you can override the allocation rule per SKU or per order value.

Lot and FEFO tracking toggle. For any SKU with an expiry or lot number, confirm this is on by default at every node, not enabled only on request. A provider that treats lot tracking as an add-on is telling you it is not core to how the warehouse runs.

Carrier rate-shopping rule. Whether the WMS selects the cheapest available carrier service per shipment or a fixed carrier per zone. Rate-shopping usually saves money in aggregate but can quietly swap a fast service for a slow one on a specific order if you have not set a minimum service-level floor.

Step 5: Verify the fit before you sign

Run a pilot before committing to a multi-year contract, and measure it against numbers you define, not numbers the provider reports back to you unverified. Pick accuracy: the rate at which an order ships with exactly the SKUs and quantities ordered, checked against your own order data, not theirs. On-time-to-cutoff rate: the share of orders received before cutoff that actually ship same day. Split-shipment rate: the share of multi-item orders that ship in more than one package, and why. Mis-ship rate: wrong item, wrong quantity or wrong address, tracked separately from carrier-caused delay.

Also verify access, not just performance. Can you pull raw inventory and order data through an API without opening a support ticket every time. Can you see which node an order shipped from, in real time, not in a monthly summary. Can you change the multi-node allocation rule yourself, or does every change require a call to your account manager. A provider that performs well during a pilot but restricts your visibility into why is asking you to trust a black box once volume climbs past what a pilot can show you.

What is the difference between a 3PL broker and a 3PL operator?

A broker manages the sales relationship and software layer while subcontracting the physical warehouse work to a partner it does not own or staff. An operator owns the building, employs the warehouse staff directly, and runs its own warehouse management system. The practical difference shows up when something goes wrong: an operator controls every lever needed to fix it, while a broker has to relay the issue to a partner whose priorities are set by its full client roster.

Which operating model fits a $3M-$30M Shopify Plus brand?

Most brands in this range with a national ship-to spread and more than a handful of SKUs are better served by a distributed operator than a single-node broker, because national shipping without multiple nodes stretches transit time and cost for roughly half the country. The exception is a brand with a concentrated regional customer base and a simple, stable SKU set, where a single-node generalist can still meet the delivery promise without paying for network coverage it does not use.

Who should not use a distributed fulfillment network yet?

A brand with a low or unpredictable order volume, a long tail of low-velocity SKUs, or a customer base concentrated in one region should not move to a distributed network yet. Spreading thin inventory across multiple nodes increases the odds of a stockout at any single node, which triggers split shipments and doubles carrier charges on those orders, erasing the transit-time benefit the network was supposed to deliver.

These operating-model decisions are, underneath the vendor language, an ops-automation problem: matching a workflow’s shape to real order and inventory data instead of a sales pitch, then keeping the settings that decide day-to-day performance under your own visibility rather than a provider’s. That is the same discipline behind Pointerflow’s ops-automation work — mapping what a system actually does against what it was sold to do, and fixing the gap before it shows up as a missed cutoff or a split shipment nobody flagged.