What follows is a frame challenge.
You are asking for two fundamentally different things:
- how to solve Sudokus "linearly"
- how to solve Sudokus "efficiently"
A restriction on the order/manner in which cells are filled in will cause a decrease in efficiency. For any given Sudoku puzzle (really, for any given grid deduction puzzle), there will be one or more "natural" solution paths. These are sequences of deductions that naturally follow from each other.
Filling in a Sudoku cell will immediately eliminate other possibilities within that box, row, column, etc. with possible knock-on effects elsewhere. Once one cell is filled, some other cells will be easier to solve. So on and so forth. While it may be technically possible to make deductions which are off of this "natural" solution path, it would typically require more involved deductions with complex chains of reasoning.
There is no guarantee that the "natural" path will be "linear" as you define it. A setter may design a puzzle such that the "linear" solve is a natural one, similarly to how some Sudokus are purposely made such that filling in all 1s, then all 2s, then all 3s, etc. is a "natural", intended path. But if you're trying to solve Sudokus in general by forcing deductions to occur in a "linear" order, that will naturally lead to a decrease in efficiency.
Thus, there are no tips I can possibly give you that will increase "efficiency" by forcing deductions to occur "linearly".