By Haru · Updated · 2026-09-07
How to Train a Dog — Four Rules Before Any Command
Most training advice is a list of commands with instructions for each. Sit like this, stay like that. It is not wrong, but it skips the part that decides whether any of it works, which is why the same person can follow ten guides and end up with a dog that sits beautifully in the kitchen and ignores them everywhere else.
Four rules do most of the work: teach one thing at a time, mark the right instant, know how and when to thin out the rewards, and expect to start again in every new place. Get those right and the commands almost train themselves. Get them wrong and no amount of repetition helps.
This page sets out those four and then applies them to the four problems people actually search for — house training, barking, pulling on the lead and jumping up at people.
One boundary before starting. Aggression, real fear and separation distress are not the subject here. Those need a qualified behavior professional working with the animal in front of them, and a web page that pretends otherwise does harm.
Rule one: one behavior, one criterion, at a time
The most common mistake is asking for several things at once without noticing. Sit, hold it, while a visitor comes in, off the lead, in a busy room. That is five criteria, and a dog that fails is not being stubborn — it is being asked a question it has not been taught.
Split it. Teach the behavior first with nothing else going on. Then add duration, then distance, then distraction, one at a time and never two together. If something falls apart, you added two.
This also tells you what to do when progress stops. Go back to the last version the dog could do reliably, run it a few times to rebuild confidence, then raise one thing by a smaller step. Trainers call this splitting rather than lumping, and it is the difference between steady progress and weeks of frustration.
A session is more useful when you know before it starts which single thing is getting harder today.
Rule two: mark the instant, not the moment you got round to it
A reward marks whatever the dog was doing when it arrived. Deliver it three seconds late and you have rewarded standing up, looking away, or wandering off — which is exactly what tends to get trained by accident.
The fix is a marker: one short sound that always means the food is coming. A word said the same way every time works, and so does a clicker. Charge it first by making the sound and feeding, twenty or thirty times, until the sound alone makes the dog look for the food.
After that the marker buys you time. You mark the instant the rear touches the floor, and the treat can take two seconds to arrive without teaching anything wrong.
This single change fixes more stalled training than any other. If a dog seems to be learning something other than what you meant, the timing is the first suspect.
Rule three: thin the rewards out, but not by stopping
Food is how the behavior gets built. It is not how it stays, and people usually get the transition backwards by stopping abruptly and concluding the dog was only doing it for the treat.
While teaching, reward every correct repetition. Once the behavior is reliable in that setting, move to rewarding unpredictably — sometimes the first, sometimes the third, occasionally two in a row. Unpredictable rewards produce the most persistent behavior, which is the same reason slot machines work.
Then widen the definition of reward. Going out of the door, being let off the lead, having the ball thrown and being greeted are all rewards, and using them keeps the behavior paid without a pocket full of food.
Expect one wobble on the way. When a reward that used to come every time stops, behavior briefly gets more intense before it settles — the dog tries harder, barks once, offers the sit twice. That burst is normal and short, and giving in during it teaches exactly the intense version.
Rule four: a new place is a new lesson
Learning does not automatically transfer. A dog that sits perfectly in the kitchen has learned to sit in the kitchen, with your body in that position, in that light, with those smells. The park is a different problem, and the dog is not defying you when it fails there.
Plan to teach it again in each new place, which goes far faster each time. Three or four locations in is usually where it starts to look like the dog simply knows it.
Distance and distraction are separate axes and should be raised separately. A dog that can hold a sit while you walk five steps away cannot necessarily hold it while another dog walks past at the same distance.
And keep sessions short. Two or three minutes, several times a day, beats twenty minutes once — attention fades, and ending while the dog still wants more is what makes it come back keen next time.
The four things people actually search for
House training is a management problem before it is a training one. Take the dog out after sleeping, after eating, after play and roughly every hour for a young puppy, reward outside within a second of finishing, and supervise or confine indoors so mistakes cannot be practiced. Cleaning accidents with an enzyme cleaner matters because ordinary cleaners leave a scent marker.
Pulling on the lead persists because pulling works — the dog gets to move forward. Change what pulling produces: stop when the lead tightens, move again when it loosens, and reward the position you want while the dog is in it rather than after it has drifted.
Jumping up is rewarded by attention, including the attention in pushing a dog down or shouting. Remove that reward by turning away, and pay heavily for four paws on the floor. Ask visitors to do the same, because one person who greets a jumping dog undoes a week.
Barking is not one problem. Barking at the window, barking for attention and barking when alone have different causes and different answers, and the last one belongs in the next section rather than here.
What not to do, and when to call someone
Punishing a growl is the single most costly mistake. The growl is the warning; suppress it and the feeling underneath does not change, so the next signal may be a bite with no warning at all. Treat a growl as information about distance and work on the cause. This is the professional position, not a style preference. The American Veterinary Society of Animal Behavior recommends reward-based methods for all dog training, and does not recommend dominance theory as a guide. Aversive methods carry welfare risks and work less reliably.
The dominance framing — that unwanted behavior is a bid for status and the answer is to assert rank — rests on old observations of captive wolves that later work did not support. It leads to confrontation as a first move, which is both unnecessary and risky.
Tools that work by hurting or startling can suppress behavior without teaching an alternative, and they carry a known risk of making fear worse. Where a behavior needs to stop, teaching what to do instead gets further and does not add fallout.
Call a qualified professional for aggression toward people or other animals, for real fear, and for distress at being left alone. Those are not stubbornness and not a training gap, and general advice is the wrong tool. Ask about the methods they use before booking, and for anything that might be pain-related, see a vet first — sudden behavior change is sometimes a medical sign.
Sources American Veterinary Society of Animal Behavior (AVSAB) — Position Statements
Frequently asked questions
When should training start?
As soon as the dog is home, whatever its age. Puppies learn from the first day and every interaction teaches something whether or not you meant it to. Older dogs learn fine too — the idea that they cannot is folklore. What changes with age is the history you are working alongside, not the ability to learn.
How long should a session be?
Two to five minutes, several times a day. Attention and enthusiasm fade quickly, and a session that ends while the dog still wants more is the one it comes back keen to. Twenty-minute sessions usually mean the last fifteen minutes are teaching the dog that training gets boring.
Do I have to use food, and will I need it forever?
Food is the easiest reward to deliver quickly and precisely, which is why it is used for teaching. You will not need it forever. Once the behavior is solid you move to rewarding unpredictably and then to real-life rewards — the door opening, the lead coming off, the ball. What you should not do is stop abruptly and conclude the dog was only in it for the treat.
My dog knows the command but ignores it outside. Why?
Because learning does not generalize on its own. The dog learned it in that room, with those smells and that little audience. Teach it again in each new place, starting easier than you think you need to, and it gets faster every time. Somewhere around the third or fourth location it starts to look like the dog simply knows it.
Is a training class worth it if I can read about it?
Often, for one reason that reading cannot supply: someone watching your timing. Almost everyone marks late at first and cannot see it in themselves. A class is also controlled practice around other dogs, which is the hardest condition to arrange at home. Ask what methods are used before you book.
What if the problem is aggression or fear?
Then this page is not the right resource, and neither is most of what you will find by searching. Aggression, genuine fear and distress at being left alone need a qualified behavior professional who can see the animal, and often a vet first, because a sudden change in behavior is sometimes pain. Delaying that while trying general advice usually makes the problem harder to fix later.