Why No Formula Can Replace This Choice
Given everything on the grid from the last chapter, it would be reasonable to expect one more step: a checklist that takes a project's specifics and outputs the right method. That checklist cannot honestly be written, and not for lack of trying.
The two-question grid describes every method precisely, but it cannot choose one for a specific project. That choice depends on what the work is actually for, information no grid, and no book, can supply in advance.
The Missing Piece No Checklist Can Supply
What a project's method actually depends on is not a property of RTK, PPK, PPP, or PPP-RTK at all. It is what that specific work is for, who will rely on it, and for how long, none of which any book can know in advance. In practice, this rarely means constant deliberation. Most work settles into one dependable default, whichever method fits a person's usual site conditions, equipment, and connectivity, and that default rarely changes from one day to the next.
Expecting this companion to end with a checklist that outputs the right method is not an unreasonable thing to want. It is simply something this grid can never provide, because the missing piece is not a missing question. It is project-specific knowledge no book could have in advance.
The One Fork Worth Making Deliberately
The choice that actually requires attention is narrower and more specific than a general "it depends" suggests: whether a given piece of work is everyday production, or is establishing control that other work will be measured against later, by people who were not there to see how it was collected. That distinction can matter down to the individual point, not just the project as a whole. The same day, the same crew, the same receiver can reasonably use one method for routine collection and a different one for the handful of points meant to serve as control, for reasons that have nothing to do with timing, connectivity, or budget.
A live method tied into a trustworthy reference network, a statewide CORS network, for instance, builds real confidence in that network and the frame it is tied to. But a single live occupation is still one computation, produced once, by one method, at one moment. That confidence in the system does not automatically extend to independently verifying that specific result. Something like OPUS or CSRS-PPP offers a different kind of confidence: a specific point, checked independently, by a separate process, often against different reference stations and products entirely. Neither is more correct in general. They answer different questions, one about the system a point is tied to, the other about that particular point itself.
Reference frame and epoch belong in this same conversation, separately again. A deliverable that needs to land in a plate-fixed national datum and one that is fine referenced globally at the current epoch are not answering the same question, regardless of which method actually produced the coordinate.
A crew spends a routine day on a construction site using its usual live network correction for every point, fast, reliable, and perfectly adequate for the work. One of those points is meant to serve as project control that a different crew will tie into next year. Does the same method that served the rest of the day well automatically serve that one point too? Not necessarily. What that specific point needs to prove, and to whom, is a separate question from what the rest of the day needed.
None of this points toward a specific method or service. It depends on what the work is actually for, and once that much is clear, the choice itself is rarely difficult.
No formula built from the two-question grid can choose the right method for a specific project, because that choice depends on what the work is actually for, not on any property of the method itself. A dependable default handles the great majority of it. The one fork worth making deliberately is recognizing when a point is establishing control others will rely on, rather than serving its own immediate purpose.
Choosing a method well is only the first step. The next chapter turns to something that holds true no matter which of these four methods was actually used.