1) Strengthen tendencies toward a behavior that the dog can pigeon-hole as a unique behavior. This can be done with either positive or negative reinforcement.
2) Once the behavior is predictably replicable, precede the behavior with an arbitrary cue. If you wish to use different modalities (visual vs. auditory vs. tactile) put them in sequence, NOT at the same time.
3) Practice the pattern using positive reinforcement.
4) Create a situation likely to fail. Apply “punishment for failure to perform a known behavior in a timely fashion” – Do it right and do it right now. You may have to use positive reinforcement to increase likelihood of response. Then repeat this process until it happens ‘right, right now’. This is a critical step and may take several shots dropping to #3 to get to another round of #4.
5) Integrate the behavior into the dog’s repertoire so that you can get it when you want it and no other behavior conflicts. This will use reinforcement and punishment (both types of each) as needed to maintain high levels of performance and reliability.
6) Maintain the behavior as needed. If it fails, go back to step 1. Do not try to take a short cut by dropping to a slightly lower level of performance. You may have a fundamental error embedded in the behavior at any part of the process.
p.s. If you can’t state a concise formula for what you do, work on it. Simplifying it to basic principles helps clarify the process. It also makes trouble-shooting easier.