Think of this example: we are researching for a few months whether an arbitrary compound X has relieffing effects on people with headache, but since we are still researching, there is no concrete result or any slight evidence yet. maybe the compound will relief the headache in 30min but maybe it will make your nose bleed.
but there is this conference about headaches that is approaching and why not submit a paper about the work with some results? WHAT RESULTS? You will spend countless hours reading (more) about headaches and compound X to write the background of the manuscript while you still have to figure out the tests you will perform test your assumption.
you will work for weeks (under more pressure) with the hope you could publish something that is very uncertain at the current state.
Deciding to publish the results can just as easily be restated as "Come up with an experiment with a clear and precise scope"
Where to publish is often determining what audience and what questions to ask. For example, targeting a clinical journal means proofs and analytical results may be less compelling, while a simulation of a study might be of active interest.
Both of those help narrow the scope of "All possible interesting questions" to something tractable.
However, unless you stumble upon something that works out very nicely right before a deadline, a publication requires planning many months in advance what you are going to do.
Because publications pretty much define your scientific contributions, i.e. your productivity as a researcher.
> One should get the results AND THEN publish it.
You must have had a body of work that was just about worthy of publication though, right?