On this page
Maya has a 9-month-old, 48-pound Lab mix who throws a perfect sit to beg for dinner, but stares blankly when she says “sit” on a walk. She is weighing the choice every owner hits: keep repeating the word louder until it sticks, or stop saying it entirely and rebuild the sequence the right way. I counsel the second path. Expectation first — the word is not what teaches the behavior, the behavior teaches the word, and you add the cue only once the movement is fluent and predictable.
That late-addition rule is my evidence-bounded recommendation for how to add a command to a trained behavior. Name the behavior you want, shape it silently with a marker and reward, then attach the verbal cue like a label after the fact. The strongest counterargument is the old-school instinct to lead with the command and lure into it from day one, which feels faster. I still land on cue-last because leading with the word trains dogs to tune it out, while cue-last keeps the word clean and meaningful.
The TikTok loop that gets this backward
Scroll training TikTok and you will see the same clip: owner says “down, down, DOWN” while pushing a puppy’s shoulders, then calls the dog stubborn. The sound trends because frustration is relatable, not because the method works. A cue said before the dog understands the movement becomes background noise, and physical prompting adds stress that slows learning.
The prevention-minded baseline I give clients comes from the AVSAB guidance on humane, reward-based training: teach the movement first with low-stress handling, then add the cue. In practice that means capturing or luring a sit five out of five times without any word, marking and rewarding each rep, before the word “sit” ever enters the room.
By day 10 of that silent work, most pet dogs offer the behavior the second they see you reach for treats. That anticipation is your green light.
When I learned to shut up first
I had a 2-year-old, 62-pound doodle who “knew” spin because the owner chanted “spinny-spin-spin” while luring with cheese every night for three weeks in the living room. On leash at the park, zero spins, just jumping and barking. The owner’s concern was real: did the dog need a stricter correction, or was the cue itself broken? We weighed staying the course with more repetitions versus stripping the cue completely.
We stripped it. For one week we rewarded any tight clockwise turn with no word at all, just a marker and a treat placed to reset. By day 7 the dog was throwing four crisp spins in a row to earn the reward, tail low and focused instead of frantic. Only then did we reintroduce a new word — “twirl” — said once, right as the shoulders started to turn. The old chant had been poisoned by repetition without understanding, and trying to fix it would have taken longer than starting the label fresh.
That tradeoff is common.
How to layer the cue without breaking it
Wait for predictability. You should be able to guess the behavior is coming before it happens.
Say the new cue once, during the behavior, then mark and reward. For sit, that means “sit” as the elbows are bending, not before the dog looks at you. Repeat for 15 to 20 short reps across two days.
Then move the cue one beat earlier. Say it as the dog starts to offer, pause two seconds, and only help with a subtle hand signal if needed. If you have to repeat the cue or escalate your tone, you moved too fast — drop back to cue-during-motion.
The American Kennel Club basic-commands guide describes the same progression for sit and down: lure the motion, reward the motion, then attach the word once the pattern is solid.
Clean cues stay quiet.

The heel case that changes the rule
A client with a 4-year-old, 35-pound cattle dog mix wanted “heel” for a narrow apartment hallway, not the park. The dog already walked close beside its owner indoors for treats, but outdoors she pulled toward bikes and pigeons. The edge case was context: the behavior was fluent in one setting and invisible in another, and the owner worried the cue would fail when it mattered most.
We did not add “heel” to the outdoor walk yet. We kept the new cue indoors where success was at 90%, added distance in five-foot increments, then rebuilt the behavior from scratch outdoors at a quiet cul-de-sac at 7 a.m. before generalizing to noon foot traffic. By week three the dog could hold five steps of close walking outside on the new cue, because we treated indoor heel and outdoor heel as two separate behaviors with the same label applied last.
If your dog only does it in the kitchen, the cue is not ready for the sidewalk.
My take for your next session
Pick one behavior your dog already offers, drop the word for three days, and reward the movement until it loops. Then add the cue once, during motion, and prove it in a new room before you trust it outside. For marker timing refreshers, see Clicker vs Verbal Marker Word: Which One Actually Trains Faster? and Can You Train a Dog Without Treats? What Actually Works. If classes would help your timing, read Are Dog Training Classes Worth It? A Vet’s Honest Breakdown.
A quiet cue added late beats a loud command repeated early, every time.
Frequently asked questions
When should I add the word to a behavior my dog already does?
When you can predict the behavior before it happens and your dog offers it repeatedly for a reward without prompting. Add the word during the motion for 15 to 20 reps, then shift it slightly earlier.
Why does my dog listen at home but ignore the cue outside?
Dogs learn in context. Fluency in the kitchen does not transfer automatically to the park. Rebuild the behavior silently in the new setting, then reattach the same cue once the movement is reliable there.
What if I already repeated the cue too many times?
Retire the poisoned word for two weeks, retrain the movement silently, then attach a fresh cue. A new, clean word attached late works faster than rehabilitating one your dog has learned to ignore.
Sources
- Position Statements — AVSAB