Shipping Orders From Your Stores: The Real Cost Per Order

Shipping from a store saves money on distance and spends it on handling. That much is uncontroversial. What is surprisingly rare is a retailer who has priced both sides on their own numbers, which is how store fulfillment ends up being decided on proximity alone and then quietly disappointing in the accounts.

This is not an argument against it. Stores hold stock close to customers, and that is a real structural advantage. It is an argument for doing the arithmetic before turning it on everywhere, and for knowing which numbers actually decide it.

What changes when a store ships the order

The parcel starts somewhere else, so it travels a different distance. The person picking it is a retail associate working between customers rather than a warehouse picker. The packing happens at whatever bench the stockroom has, with whatever boxes have not run out.

Those are the three that get modelled. Several more rarely do: a carrier now has to collect from every participating location rather than one dock, packaging consumables need replenishing at every store, returns need a route back, each store needs a printer, a scale and someone who knows what to do when they fail, and multi-line orders a store cannot fill whole turn into splits with a second label. None of these is large on its own. Together they are usually bigger than the line item anyone argued about.

The saving everyone leads with: distance

Carrier pricing is banded by distance. The Postal Service explains that its zone charts determine "the distance that a mailpiece travels based on the ZIP Code where the mailpiece enters the mailstream (origination) and the ZIP Code that will deliver the mailpiece to the addressee (destination)", with zones "designated as local or 1 through 9". Private carriers use their own tables built on the same idea.

Two things follow. A store forty miles from the customer prices in a lower band than a warehouse nine hundred miles away. And because the bands are built from ZIP code pairs rather than drawn as circles, the boundaries are irregular, so the only way to know what a move is worth is to price real addresses from both origins on your own rate card.

There is a second and often larger effect hiding behind the first. Distance sets not just the price of a service but which service can hit a promised date. A parcel that has to cross the country in two days may need an air product; the same promise from a store thirty miles away can often be met on ground. Moving down a service level usually saves more than moving down a band, and it is the effect most worth checking first.

What none of this tells you is the amount. Published bands are not prices, the lowest bands are often priced very close together, and any retailer with thirty stores is on a negotiated card anyway. Anyone who quotes you a per-order saving without opening your rate card is guessing.

The cost almost nobody prices: the minute

A store pick is done by a retail associate. The Bureau of Labor Statistics puts median pay for retail sales workers at 17.10 dollars an hour as of May 2025. At that rate a minute of base wage is about 28.5 cents, so a ten minute fulfillment task is about 2.85 dollars.

Treat that as a floor rather than an answer, for two reasons.

The first is that base wage is not what an hour costs you. Payroll taxes, benefits and paid time off put a loaded rate well above the base, so the real figure for those ten minutes is meaningfully higher. Use your own loaded rate.

The second matters more and is usually missed entirely. What you want is the marginal cost of the pick, not the average. If the associate was already scheduled and the store is quiet, those minutes were being paid for regardless and the true incremental cost is close to nothing. If the store is busy, the cost is not the wage at all: it is the sale that did not get served while they were in the stockroom, which is worth considerably more than 28.5 cents a minute. The same ten minutes can be nearly free at eleven on a Tuesday and the most expensive minutes of your week on a Saturday afternoon.

That is the real finding, and it is the one the business cases skip. Store fulfillment is not uniformly cheap or expensive. It is cheap in slack hours and expensive in peak hours, and a routing rule that ignores the clock is leaving the largest variable in the model unmanaged.

The parcel does not get cheaper because it left a store

Carriers do not bill on weight alone. FedEx states it directly: "For each shipment, you are charged based on the dimensional weight or actual weight of the package—whichever is greater", where dimensional weight is length times width times height in inches divided by 139 for US shipments. UPS publishes the same calculation using 139 for daily rates and 166 for retail rates, rounding any fraction up to the next whole pound. Both are list values; your contract may differ.

A box measuring twelve inches on each outside edge is 1,728 cubic inches, which divided by 139 is 12.43, billed as 13 pounds unless the contents weigh more, in which case the scale weight wins. The same goods in a box measuring ten inches outside are 1,000 cubic inches, or 7.19, billed as 8 pounds. Five billable pounds, decided by which box someone reached for. What those five pounds are worth again depends on band and contract, so price it on your own card rather than assuming.

The store-specific problem here is not usually carton range. Most chains running store fulfillment issue a standard kit of four to six sizes specified by the distribution center. The problem is replenishment and ownership: the right size runs out, nobody in the store is accountable for reordering it, and the associate ships the item in a box that is too big or in its own retail packaging. That is a supply chain fix, not a purchasing decision, and it will not show up anywhere except on the invoice.

What it does to the store

The store is now running two operations in one room with one set of people, and the pick competes with the customer standing in front of the associate. That tension is manageable with rules about when stores receive work, and unmanageable without them.

It also rests on the store's stock record being right, because the order was promised against that number. When the unit is not there, a well configured system does not cancel: the store rejects the line and it reassigns to another location or the distribution center. The cost is a delay, a second pick and sometimes a split, which is real but recoverable. The two numbers worth watching are the reject rate, which is usually highest at rollout and falls as stores learn the process, and the time to reject, because a line sitting unanswered in a store for six hours is what turns a recoverable miss into a late delivery. Neither improves on its own, and both depend on measured on-hand accuracy rather than on how confident a store manager feels.

This is why one reliable stock position across locations is a prerequisite rather than a later refinement, and why which location should ship a given order has to be answered with rules rather than instinct.

When shipping from the store wins

It wins when the distance or service-level saving is large and the handling penalty is small: bulky but light items where dimensional weight already dominates and the store can pack them properly, items only stocked in stores, addresses far from the distribution center but near a store, end-of-season stock where shipping beats marking it down, and orders the store can pick in slack hours. It also wins when the alternative is telling the customer no.

When the warehouse should still win

Small dense items where the band saving is a few cents. Multi-line orders a store would have to split. The store's peak trading hours, when the minute costs a sale rather than a wage. Anything needing packaging or handling the store does not have. A distribution center exists to do this work cheaply, and routing work to it is not a failure of the store network.

How to decide, in order

1. Price the service level first. Check whether store proximity lets a cheaper service meet the same promised date, since that is usually the largest single effect.
2. Price the distance saving on your own rate card, for addresses you actually ship to, from both origins.
3. Time the pick in a busy store and in a quiet one, and cost both at your loaded rate. Keep the two numbers separate; they are different decisions.
4. Compare billed weight to scale weight by origin over a month, then fix consumable replenishment before widening any program.
5. Watch reject rate and time to reject from the first week, and hold stores to a measured accuracy figure.
6. Turn it on only for the cases these numbers already justify, and re-check when rates or wages change.

Store fulfillment is a tool for specific situations rather than a default setting. The work of finding those situations is a pilot measured over a quarter, not an afternoon with a spreadsheet, and omnichannel fulfillment pays when each order goes to the place that can actually serve it most cheaply, which is not always the nearest one.