One carrier is a single point of failure
The advice to pick one carrier and negotiate hard made sense when integrating a second one was a project.
Consolidating volume with one carrier buys a better rate. It also buys a correlated risk: when that carrier has a bad month, every one of your parcels has a bad month, and you have no lever to pull.
What a mix is actually for
The argument for multiple carriers is not primarily price, although price is usually part of it. It is that different carriers are good at different things, and those differences are stable enough to route on.
- Domestic density: the carrier with the most drivers in your country wins on cost.
- Cross-border lanes: a regional specialist frequently beats a global network on its own routes.
- Collection points: lockers and shops are a different network from home delivery.
- Heavy and oversized: past the parcel ceiling it is a freight decision, not a parcel one.
The objection, and the answer
The reason shops stayed single-carrier was integration cost. Every additional carrier meant another API, another label format, another tracking webhook and another set of edge cases to maintain. That was a real reason.
It stops being a reason when the integration is done once, at the layer above the carriers. At that point adding a carrier is a configuration change, and the question returns to the one that should have been asked all along: which service is right for this parcel?