"Is your method a sort of rule-of-thumb or a standard practice?"

Neither. Its a question of divide and rule. For a typical position control loop this is one way to tune :

You start with only the P-Term. Keep changing the Set Value (SV) and see how best the Process Value (PV) follows your command. If PV overshoots - reduce P-Term, if PV undershoots or does not catch up fast enough then increase P-Term. Once you have optimized this then lock P-term.

Do you notice that the PV never really comes up to the SV when you stop changing SV? Then its time to bring in the I-Term to knock out the steady state error. Start with something like a 10% setting and adjust till PV=SV under steady state conditions.

OK the D-Term is tricky - it monitors the rate at which the PV is "approaching" the SV term and can help to smoothen out the overshoots even when a high P-Term is used.

End of lecture.

The above is the verbatim of an algorithm that we developed for self tuning of a hydraulic servo positioner for a training kit. Very basic but it did deliver the goods. And if you want the analog Op-Amp circuit that we used its available . Mail me.

Raghu