In-house or 3PL is not a cost question

The spreadsheet is always the first artefact. In-house on the left. 3PL on the right. Cost per order, cost per pick, cost per return. Someone adds a line for "flexibility" that nobody can number, then deletes it because it looks unprofessional. The cheaper column wins. The meeting ends.

Nine months later the cheaper column is still cheaper, and the operation is a mess you cannot see. Or the expensive column is expensive, and you still cannot see it, because you outsourced the thinking along with the labour.

I've built an operation we ran ourselves. I've sat inside a 3PL running other people's operations. The cost question is real. It is also the wrong first question.

In-house or 3PL is not a cost question. It's a question about which parts of the operation you are willing to not understand.

Cost is the easy column

Unit cost is a comforting number. It sits still on a slide. It can be compared. Procurement knows what to do with it. Finance can put it in a model that assumes volume arrives in a straight line and exceptions stay at last quarter's rate.

Operations does not sit still. The thing that blows up the model is not the pick fee. It's the failed delivery loop, the peak week the 3PL "can flex" until they cannot, the SKU that needs a special pack, the returns that come back to the wrong node, and the fact that your customer service team is now arguing with a partner's customer service team about a status neither of you trust.

A 3PL that looks cheaper at 500 orders a day can get very expensive at 5,000, not because they changed the rate card, but because you never priced the exceptions, the management time, or the days you spend flying to a warehouse you do not control to find out why the dashboard is green and the customers are not.

What you are actually choosing

In-house means you own the process, the people, the floor, and the 2am problem. That is heavy. It is also honest. When something breaks, you can walk over to it. You can change a pack bench without a change request. You can see whether "despatched" means on a truck or in a tote waiting for a truck.

3PL means you are buying capacity, labour, and (if you are lucky) a process they already run well for someone who looks a bit like you. You are not buying a brain. You are not buying accountability that survives a QBR. You are buying execution of a brief. If the brief is thin, the execution will be faithfully thin.

I've watched brands treat a 3PL as a way to postpone becoming operators. They hand over cartons and a launch date. They keep the P&L. They lose the ability to explain, in a sentence, how an order becomes a delivery. Then they are surprised when the partner optimises for the contract, not for the customer.

That is not a 3PL being evil. That is a 3PL being a 3PL. Their job is to run what you bought. Your job is to know what you bought.

The control you think you kept

People say they outsource fulfilment and keep "strategy." Strategy is a slide. Fulfilment is where the brand actually happens. If a customer in Jakarta gets a crushed box, they do not care that your strategy was asset-light. They care that you sent them a crushed box.

The control that matters is not owning the lease. It's owning the definition. What does cut-off mean. Who can split a shipment. What happens when COD fails. Who decides to hold a line because the commercial invoice will bounce. If those definitions live only in a partner's head, you do not have an operation. You have a dependency.

In-house teams fail the other way. They confuse ownership with competence. Having your own warehouse does not mean you have a process. It means you have rent, and a night shift, and a very expensive place to hide a mess. I have seen in-house floors that were less visible than a decent 3PL, because nobody wanted to admit the SOP was a rumour.

You can outsource labour. You cannot outsource knowing how the work actually runs.

How to ask it properly

Start with the shape of the work, not the rate. How messy are the SKUs. How peaky is the calendar. How many markets. How ugly are the exceptions. How much of this is a process you could brief tomorrow versus a process you are still discovering.

If the work is stable, repeatable, and already written down, a 3PL can be a very good buy. You are purchasing a machine you understand. If the work is still being invented, a 3PL will freeze the invention at whatever you happened to brief in week one, then run it faithfully while you complain.

If you need to learn the operation, keep it close until you can describe it. Then decide. Hybrid is allowed. In-house for the ugly, new, or brand-critical part. 3PL for the part you can already draw on a whiteboard. That is not indecision. That is sequence.

The goal was never a cheaper pick

The goal was an operation you can still explain when volume doubles, when a market is added, and when the person who "owns the 3PL relationship" is on a plane.

Run the cost model. Of course run it. Then put it second. First, write down which parts of this job you need to see with your own eyes. The column that wins is the one that leaves you with a process, not just a price.