Small orbital changes require orders of magnitude less delta-V than a dedicated launch, and launches have to deal with the atmosphere on top of this.

Rockets become more efficient as they get larger, you get lower structure to payload weight ratios and you suffer fewer losses to drag in the atmosphere. The same scaling goes for your fuel tanks. You want to get the biggest fuel tank you can in orbit and then use efficient ion engines to move for whatever delta-V corrections you need.

Doing some back of the envelope calculations, a single 20,000 kg depot could provide fuel at approximately 10% of the cost of individual launches even with an extraordinarily inefficient 300 m/s of delta-V change for every refueling. Realistically you could probably get under 1% with efficient planning.

Certain heavily-populated classes of orbits (like SSO, which this one is using) are very cheap in propellant terms to transfer between. Only as long as you don't care about speed of the transfer, but refueling needs are predictable very far in advance

No idea about the feasibility of this, but it seems like hitching a cheap rideshare to a non-ideal but heavily populated/serviced orbit on a bulk carrier like Falcon 9 or Starship and then contracting with a little ferry to help your sat get to its final orbit could help SpaceX negate one of the big advantages of the companies launching smaller rockets that launch direct to final orbit?

Kind of like remaking the old hub and spoke model, with Starship being the bulk 747, similar to the era when air travel was prohibitively expensive and demand was still too low to support direct point to point everywhere.