In-house fulfilment or a 3PL: how to decide

The comparison everyone runs, and why it misleads

The usual way to make this decision is to build a spreadsheet. Cost per order in-house, cost per order outsourced, pick the smaller number. It is a reasonable instinct and it produces a bad decision surprisingly often.

Two reasons. First, the in-house number is almost always wrong, because the labour is buried in salaries of people who do other things too, and nobody costs the founder's Sunday. Second, and more importantly, the two options do not fail in the same way, and the failure mode matters more than the unit cost at the volumes most stores operate at.

So run the spreadsheet, but do not decide from it. Decide from what each option does to your week when things go wrong.

What in-house actually costs

To get an honest number, count these.

• Labour, at a real hourly rate, including the hours worked by people whose job title says something else.
• Space, including the portion of your unit or garage or shop back-room that stock occupies and could otherwise be doing something.
• Packaging and consumables, which are easy to count and usually already counted.
• Software and hardware. Printers, scanners, whatever shipping tool you use.
• Shipping rates you can actually get. This is often the biggest single line, and the one where a 3PL's volume genuinely beats you.
• The cost of the constraint. If dispatch stops when one person is ill, that is a cost even in months when nobody is ill.

The last one is the one stores forget and the one that eventually forces the decision.

What a 3PL actually costs

Outsourced pricing is rarely a single number, and the headline pick-and-pack fee is not where the surprises are.

• Receiving, per pallet or per carton or per unit, and it varies a lot.
• Storage, per pallet or per bin per month. Slow-moving stock is expensive to leave with somebody else.
• Pick and pack, usually a base plus a per-additional-item rate.
• Shipping, at their rates, sometimes with a margin on top.
• Special handling, inserts, gift notes, kitting, anything non-standard.
• Returns processing, which is easy to overlook and, for some categories, substantial.
• Minimums. Monthly minimums make a quiet January more expensive than it looks.

Ask for a quote modelled on your actual last three months of orders, not on an average. Averages hide the shape, and the shape is what you pay for.

The question that decides it

Once both numbers are honest, they are often closer than expected, and the decision comes down to something else: which of these two problems would you rather have?

In-house fails as a capacity problem. Volume spikes, or somebody is off, and orders sit. You can see exactly what is wrong, you have complete control over fixing it, and fixing it means somebody working late.

A 3PL fails as a visibility and leverage problem. Something goes wrong in a building you cannot enter, you find out from a customer, and fixing it means an email to an account manager and waiting.

Neither is worse in the abstract. But they suit different people and different businesses, and knowing which one you can tolerate is more predictive of satisfaction than the per-order cost.

When in-house is clearly right

Some situations point one way strongly enough that the spreadsheet is a formality.

Low volume with high value. A handful of orders a day where each one matters and the margin is comfortable. Outsourcing adds cost and removes control for very little gain.

The unboxing is part of the product. If presentation, handwritten notes or careful assembly are a real part of why people buy, a standard 3PL will not do it and a 3PL that will do it charges accordingly.

Constant SKU change. Short runs, samples, made-to-order. Warehouses are efficient at stable catalogues and awkward with churn.

You are still learning what fulfilment costs you. There is a real argument for doing it yourself for a while simply to know the numbers, so that you can evaluate quotes properly later.

When a 3PL is clearly right

Fulfilment is the constraint on growth. If the honest answer to "why did we not do more last month" is packing capacity, the decision has already been made for you.

The volume is genuinely seasonal. A store that does most of its year in eight weeks cannot staff to peak and cannot staff to trough. A 3PL absorbs that shape for a fee, which is precisely what the fee is for.

You need to ship from more than one place. Multiple regions, multiple countries, faster delivery promises. Building that yourself is a different business.

The person doing it should be doing something else. If the founder packs boxes, the cost is not the packing hours, it is whatever those hours would otherwise have produced.

The hybrid nobody talks about

The choice is presented as binary and rarely is. Plenty of stores run both, deliberately.

Common splits: bulk and predictable SKUs go to the 3PL while fragile or personalised items stay in-house; one region outsourced and the home market kept; or the 3PL runs the ordinary weeks while peak overflow is handled internally, or the reverse.

Hybrid costs more per order and buys you optionality plus a fallback. For a store that has been burned by a single point of failure, that trade is often worth it. The overhead is real though: two processes, two sets of stock, and a rule for which orders go where that has to be unambiguous.

What changes for your team either way

This is the part that rarely makes the spreadsheet, and it is the part your team will feel.

In-house means the fulfilment conversation is a physical one. Someone shouts across a room, problems are seen rather than reported, and the feedback loop is instant. It also means the work is visible to everyone, including when it is going badly.

Outsourced means the conversation becomes written, asynchronous and partial. Nobody sees a problem; somebody reads about one. The team's job shifts from doing fulfilment to monitoring it, which is a genuinely different skill and one people are not automatically good at.

Teams that move to a 3PL and stay unhappy are usually the ones that did not make that shift deliberately. Nobody was given the job of watching, so watching happened only when a customer complained. Deciding who owns the daily exception check on day one, before the first problem, is most of the difference. There is more on the mechanics of that in working with a 3PL.

Switching costs, in both directions

Whichever way you go, the move itself has a cost that rarely appears in the comparison, and it is worth putting a number on before deciding rather than discovering it afterwards.

Going out means a physical stock transfer, a period where your inventory numbers are unreliable because stock exists in two places, an integration to set up and test, and a learning curve at the warehouse where your first few weeks will be your worst few weeks. Most stores underestimate the last one. A warehouse that has never handled your products will make mistakes with them for a while, and the mistakes land during the period when you are least set up to catch them.

Coming back in is worse, because you are rebuilding a capability you dismantled: space, equipment, shipping accounts at rates you no longer qualify for, and people who know how to pack your products. Stores that outsource and then return usually describe the return as harder than the original move.

The practical implication is not that outsourcing is a trap. It is that the decision deserves more care than a per-order comparison, and that a trial with a subset of SKUs is a genuinely useful way to buy information before committing the whole catalogue.

The three months after you move

If you do outsource, the first quarter determines whether the arrangement works, and most of what determines it is set up in the first fortnight.

Decide who watches, and give them a named daily slot rather than an intention. Agree what the warehouse tells you without being asked, in writing, before the first problem rather than after it. And expect the accuracy figures to be worse at the start than they will be at month six, so you do not conclude from week two that the provider is bad.

Also plan for the questions your own team will suddenly be unable to answer. Support could previously walk to a shelf and look. Now they raise a query and wait, and if nobody has told them what the expected response time is, they will guess, and their guess will be told to a customer.

How to decide without a six-month project

Three steps, and none needs a consultant.

Get your real in-house cost per order. One month, honestly counted, including hours from people whose job is something else.

Get two quotes modelled on your actual order history. Not an average month. Your real last three months, with your real SKU mix.

Write down which failure mode you would rather manage, before you look at either number. Then look. If the numbers are close, the failure mode answers it. If one number is dramatically better, ask what it is not counting, because it usually is not counting something.

Common questions