Training a puppy is not complicated, but it is unforgiving of bad timing. Almost every failure people describe as stubbornness is a mechanical failure: the reward arrived late, the step was too big, the cue was added too early, or the behaviour was only ever practised in one room.
The most common question about how to train a puppy is which method to use. The method matters far less than the mechanics underneath it, and the mechanics are the same whether you use a clicker, a word, a piece of cheese or a tennis ball.
A puppy learns from consequences, and a consequence only teaches if the puppy can connect it to what it just did.
The window is short. Estimates vary with the task and the individual, but as a working rule, reinforcement needs to arrive within about half a second to a second of the behaviour for the association to be reliable. Anything slower and the puppy is as likely to associate the reward with whatever it happened to be doing when the food appeared β sitting down, looking away, sniffing the floor.
This is why people who train by saying "good boy" and then walking to the kitchen to fetch a treat produce a puppy that has learned that the kitchen is interesting. The praise was on time; the food was four seconds late and therefore attached to the wrong moment.
Two things solve it. Deliver the reward fast enough that no other behaviour intervenes, or use a marker to bridge the gap. Most people need both.
A marker is a signal that always, without exception, predicts a reward. It is also called a bridge, because it spans the delay between the behaviour and the food.
The mechanism is straightforward once you see it as a two-stage chain. The marker fires at the exact instant of the correct behaviour, which is a thing a human can do accurately. The reward follows a second or two later, which is a thing a human needs time for. The puppy learns the chain backwards: marker means food, and because the marker was precise, the food teaches the behaviour that was happening at the moment it fired.
Before any of that works, the marker has to be charged. Ten to twenty repetitions of marker then treat, with no behaviour required, in a quiet room, over a couple of days. A clicker is the classic choice because it is sharp, consistent and emotionally neutral. A short word β "yes" is the usual one, said flatly rather than encouragingly β works just as well and is always available.
The disadvantage of a clicker is that it needs a hand and gets left at home. The disadvantage of a word is that people use it in conversation, which dilutes it. Pick one, use it only for training, and if you use a word, keep the tone identical every time β a marker is a signal, not praise.
Dogs are not equally reinforced by everything, and the value of a reward is a property of the individual dog rather than of the item.
The useful mental model is a hierarchy. At the bottom sits ordinary kibble, which most puppies will work for in a quiet kitchen. In the middle sit small pieces of something soft and smelly. At the top sit the things that produce visible excitement β cheese, cooked chicken, a tug toy for some dogs, a ball for others.
What matters is matching the value to the difficulty. A puppy learning to sit in a quiet kitchen will work for kibble. The same puppy learning to come away from a interesting smell in a park will not, and the training will fail for a reason that has nothing to do with obedience. High-value rewards are for hard tasks and for recalls, not for everything, or they stop being high-value.
The practical tool that makes this affordable is the daily ration. Measure the puppy's food in the morning and use half of it as training rewards through the day. That keeps the calorie total where it should be, gives you fifty or sixty training opportunities, and makes the puppy work for food it was going to eat anyway.
One caution. A reward that the puppy will not eat outdoors is not a reward outdoors. If the puppy takes chicken in the kitchen and ignores it in the garden, the garden is currently too distracting for food-based training and the answer is distance, not a better treat.
Every behaviour a puppy has to learn is a chain of smaller pieces, and the commonest training error is asking for the whole chain at once.
Sitting, for a puppy, involves noticing the handler, stopping, shifting weight back, lowering the rear, holding still. If you only reward the finished product, the puppy fails six times before it succeeds once, and failure is not neutral β it is information that this activity is hard and not worth the effort.
The fix is to split the step until the puppy is succeeding most of the time. A working criterion is somewhere around eight successes out of ten. Below that, the step is too big; well above it, the puppy is ready to make it harder.
This is where luring needs care. Holding a treat to the puppy's nose and moving it backwards to fold the puppy into a sit is a fast way to get the first repetition, and a fast way to build a dog that only sits when a hand is in front of its face. The lure should be faded within a handful of repetitions: same hand movement, no food in it, then the hand movement reduced to a small gesture, then the gesture reduced to nothing.
Shaping β waiting for the puppy to offer something close to the target and rewarding the closest attempt β is slower per repetition and produces a dog that thinks. For most pet training a mixture works: lure to get the picture, then shape and raise criteria to get reliability.
Puppies have short attention spans and poor frustration tolerance, which makes length the wrong variable to increase.
A useful structure is three to five minutes, four to six times a day. That is more total training than a single half-hour session, and it fits the way a puppy actually learns: short bursts, frequent repetition, and no opportunity to become bored or over-aroused.
Each session should have one objective and should end while the puppy is still succeeding. Ending on a failure β the last repetition didn't work so we stopped β is the pattern that produces a puppy that disengages when it sees the treat pouch.
Between sessions, use the day. Ask for a sit before the food bowl goes down, a wait before the door opens, a name response before the lead goes on. This is not extra training time; it is the same training distributed into moments that already exist, and it is how behaviours become habits rather than tricks.
Two rules about when not to train. Do not train a puppy that is over-tired, because an over-tired puppy has no impulse control and the session becomes a lesson in failure for both of you. And do not train when you are frustrated, because frustration produces faster repetitions of a cue that is not working, which is the single most effective way to teach a puppy to ignore a word.
A cue is the word or signal that tells the puppy which behaviour will be rewarded. The order matters: behaviour first, then cue, then reward. Not cue, then hope.
The sequence is: get the behaviour reliably, then say the word just before the puppy does it, then reward. Repeat until the word predicts the behaviour. Only then start using the word to ask for it.
The mistake that ruins cues is repetition under pressure. Saying "sit, sit, sit, SIT" teaches a puppy that the first three words are noise and the fourth is the one that matters β and if the puppy does not sit on the fourth, the word has been taught to mean "nothing happens".
Say it once. If the puppy does not respond, the behaviour is not yet reliable under those conditions, and the answer is to make it easier rather than to ask again louder. A cue that has been used a hundred times without a consequence is a broken cue, and repairing one takes longer than teaching a new word.
The other failure mode is the poisoned cue. If a word is reliably followed by something the puppy dislikes β being grabbed, being pulled away from something interesting, being put in a bath β the word acquires the dislike. This is why "come" is the most commonly poisoned cue in dog training and why many trainers teach a fresh recall word rather than trying to repair a damaged one.
Distance, duration and distraction are the three ways a behaviour can get harder. They should be increased one at a time, and slowly, because increasing two together produces a failure that tells you nothing about which one caused it.
A sit that is reliable at one metre in the kitchen is not reliable at five metres, and neither is reliable in the garden. The progression is: same behaviour, same distance, new place. Then new distance in the old place. Then duration. Then a mild distraction. Then distance with a distraction.
People skip straight to the park because that is where the behaviour needs to work, and the result is a puppy that fails at everything simultaneously and learns that the park is where cues stop working.
A rough guide to how fast to move: if the puppy succeeds eight or nine times out of ten, the next increment is safe. If it succeeds half the time, you have moved too far and should go back one step for a few sessions. The behaviour does not degrade from repeating easy steps; it degrades from repeating failures.
A dog does not learn "sit". It learns "sit in this kitchen, on this floor, facing this person, at this time of day".
This is not stupidity; it is efficient learning. An animal that generalised from one example would be constantly wrong about the world. The consequence for training is that every behaviour has to be taught again, cheaply, in a handful of new places before it becomes portable.
Five locations is a reasonable target before treating a behaviour as learned: two rooms in the house, the garden, the street outside, and one novel place such as a friend's kitchen or a quiet car park. Each new place will look like a regression, and it is not β it is the first repetition in that context.
The same applies to people. A puppy that sits for one family member has learned to sit for that person. Everyone in the household should run the same cue, with the same criteria, or the puppy will learn that the rules are person-specific β which, from the puppy's point of view, is an accurate reading of the evidence.
Puppies sleep somewhere between sixteen and twenty hours a day, in scattered bouts, and a puppy that has not slept is a puppy that cannot learn.
The behavioural signature of an over-tired puppy is the same as the signature of a disobedient one: biting, zooming, ignoring cues, throwing itself at things. Owners read this as excess energy and respond with more exercise and more training, which produces a more tired puppy behaving worse. The correct response is a dark quiet room and a sleep.
A structured day helps. Puppies do well on a rhythm: awake for an hour or so including a training session and a toilet trip, then asleep for two. Building that rhythm early prevents most of the late-evening biting that owners describe as a personality change.
There is a developmental period, roughly from three weeks to around sixteen weeks, during which a puppy forms its durable impressions of what is normal. Experiences in that window β people, sounds, surfaces, other animals, being handled β shape the adult dog's default response to novelty for the rest of its life.
After roughly sixteen weeks, the same experiences are still learnable but the fear response is harder to overwrite, which is why a dog that never met a stranger until six months often stays wary.
The objection people raise is vaccination: how do you socialise a puppy before its course is complete? The answer used by veterinary behaviour bodies is that socialisation and infection risk are both real and can be managed simultaneously rather than traded off. Carry the puppy in public rather than letting it walk everywhere. Meet people and calm, vaccinated, healthy dogs in clean environments β a friend's living room, a quiet garden. Avoid high-traffic dog areas, parks where unknown dogs congregate, and any surface heavily used by unfamiliar dogs.
What is not acceptable is a puppy kept indoors until sixteen weeks and then expected to be confident, because the window does not wait for the vaccination schedule.
Somewhere between six and eighteen months, depending on breed and size, a trained puppy becomes unreliable. Recall fails, the door-bolting returns, previously solid cues are ignored, and the dog that came when called now looks at you and considers its options.
This is normal and it is developmental. Adolescence brings increased independence, a reorganisation of social priorities, and a second fear period somewhere around eight to ten months. It coincides with the point at which owners relax, which is why so many dogs lose their training in the second year.
The response is not to start again from zero; it is to keep the reinforcement rate up. A dog in adolescence needs more reward for compliance than it did as a puppy, not less, because the competition for its attention has increased. Short daily sessions, high-value rewards for the things that matter β recall above all β and a return to easy versions in new places will carry most dogs through.
The behaviours that survive adolescence best are the ones that were generalised early, reinforced often, and never punished. A recall that is sometimes punished is a recall that does not survive.
This is a common and specific problem, and it is a clue rather than a training failure.
A puppy that takes food happily indoors and refuses it outdoors is telling you that the outdoor environment is above its threshold. Refusing food is one of the earliest signs of stress in a dog, and it appears before the more obvious ones.
The sequence that works is distance. Find the point at which the puppy will still eat β often twenty metres from the trigger, sometimes the far side of a car park β and train there. Then close the distance over sessions. This is called working under threshold, and it applies to every dog that finds the world too much rather than too little.
What does not work is a better treat at the same distance. A puppy over threshold is not evaluating flavour; its digestive system has been told to stand down. The treat is not being rejected because it is boring.
Recall is the behaviour that keeps a dog alive, and it is also the one most often damaged by ordinary handling.
The mechanism is unforgiving in one specific way: a recall that is punished once is a recall that degrades. "Punished" does not mean hit. It means anything the dog finds unpleasant happening after it arrives β being put back on the lead to go home, having a toy taken away, being grabbed for a bath, being shouted at for taking too long. Each of those teaches the dog that coming to you is the moment the fun stops, and dogs learn that lesson in a single repetition.
The way to protect it is to keep the reinforcement ratio absurdly high. Reward most recalls, for the dog's whole life, and make the reward better than whatever the dog was doing. Coming away from a smell requires paying more than the smell is worth, which in practice means real food or a game, not praise.
Three structural habits help. First, practise a recall you can always win: in the house, in the garden, on a long line, at short distances, with nothing competing. Second, never call a dog for something it will dislike when you can walk to it instead β go and get the dog rather than calling it to the bath. Third, teach an emergency word that is only ever used for a jackpot reward, so that a rare and expensive treat predicts the one cue you cannot afford to have ignored.
The long line is the tool that makes this safe. A ten-metre line on a harness lets a puppy rehearse recall in the open with a physical guarantee, and it is the difference between practising recall and hoping for it. Retractable leads are not a substitute β they apply constant tension, which teaches a dog to pull against pressure, and they cannot be used to stop a dog.
Some of the most valuable training a puppy receives does not look like training at all, because it is the set of behaviours that make veterinary care and daily life possible.
Being handled. Touch the ears, lift the lips, hold a paw, run a hand along the spine β briefly, then reward. Repeat in short sessions until each is unremarkable. A dog that accepts examination is a dog that can be treated without sedation, which is worth more over a lifetime than any trick.
Nail trims. One nail, then reward, then stop. Build to two, then a paw. The target is a dog that tolerates a full trim in a few weeks rather than a dog that is wrestled once a month.
Muzzle training. A muzzle is a piece of safety equipment and every dog should accept one, because the day it is needed is not the day to introduce it. Ten seconds of muzzle on, then food through the front, then off. Build to a minute.
Lead manners. A puppy that pulls is not being dominant; it is being reinforced by arriving at things. The mechanical answer is to stop moving when the lead tightens and move again when it slackens, which teaches that a loose lead is what produces progress. This is slow and it works, and it is much faster at four months than at eighteen.
Settling. Rewarding a puppy for lying down and doing nothing is the most underrated skill on the list. A dog that can settle is a dog that can go to a pub, a cafΓ© and a friend's house, which is most of what people actually want from training.
Training has numbers, and looking at them prevents the common situation where an owner practises for weeks and cannot say whether anything improved.
Success rate. Out of ten attempts, how many worked? Under six means the step is too hard. Eight or nine means the puppy is ready for the next increment. This is the single most useful number in puppy training and almost nobody counts it.
Latency. How long between the cue and the behaviour? A sit that arrives in half a second is learned; a sit that arrives after four seconds of sniffing is not, even though it happens. Latency is the measure that catches the dog that complies eventually.
Rate of reinforcement. How many rewards per minute are being delivered? In early learning it should be high, because a puppy that is being paid frequently stays in the game. A session with three rewards in five minutes is a session where the puppy spent most of its time not being right.
Duration of engagement. How long before the puppy disengages? If it is under a minute, the session is too long, too hard or too boring, and the fix is one of those three rather than more discipline.
Counting takes a week and then becomes automatic. It converts "he's not listening" into "he's at four out of ten on this step in this location", which is a problem with an obvious solution.
A second dog changes the practicalities of training far more than most people expect, and the problems are almost all about rehearsal rather than about ability.
The first rule is separate sessions. A puppy trained alongside an adult dog is a puppy watching an adult get rewarded for behaviour it did not perform, which is confusing at best and, at worst, teaches the puppy to offer whatever the adult is offering at the wrong time. Train one dog at a time, with the other behind a door or with another person, and keep the sessions short enough that the waiting dog does not rehearse complaining.
The second rule is that the older dog is a trainer whether you like it or not. Puppies copy. If the adult jumps at the door, the puppy will jump at the door, and it will do so faster than it learns anything from you. This is the strongest argument for fixing the older dog's manners first, because a puppy's undesirable habits are much easier to prevent than to remove.
The third is resource management rather than sharing lessons. Feeding side by side and expecting a puppy to learn to share is not training; it is an experiment with an unpredictable result. Feed in separate places, give chews in separate places, and let the relationship settle before testing it. A puppy that learns early that food can be taken is a puppy that learns to guard.
The fourth is that some things are worth teaching together, deliberately and once. A group settle β both dogs lying down while the household does something else β is easier to build from two separate settles than from scratch, and it is the behaviour that makes a two-dog household liveable.
The fifth is a warning about cumulative reinforcement. Two dogs means twice the barking at the doorbell, twice the pulling, and twice the noise, and each dog is being reinforced by the other's behaviour. The household that ignores a pair of barking dogs for a month has built a two-dog barking problem that neither dog would have had alone.
Training only in one place. The most common, and the one that produces the dog that is perfect in the kitchen and absent in the park.
Repeating cues. Teaches the puppy that the first three are optional.
Long sessions. Produce fatigue, frustration and a puppy that disengages when the equipment comes out.
Rewarding the wrong moment. Late food, or food delivered while the puppy is already moving, attaches the reward to the movement.
Punishing a failed recall. Ends the recall. A dog that is punished for coming learns not to come, and it learns it in one repetition.
Three to five minutes for a puppy, four to six times a day. Short and frequent beats long and occasional, and the total time is what accumulates.
Yes, and it is the cheapest method available. Measure the daily ration and use half of it for rewards through the day. It controls calories and gives you dozens of repetitions.
Either. A clicker is more precise and easier to keep consistent; a word is always available. What matters is that the marker is charged first and used identically every time.
Because it learned to sit in one place rather than to sit. Generalisation has to be taught in each new location, and the outside world also adds distraction, which is a second variable. Handle them one at a time.
The day the puppy comes home, with rewards and no corrections. What a puppy cannot do is concentrate for long, so the sessions start at a minute and grow.
No. Training works at any age. What five months costs is part of the socialisation window, which makes novelty harder to accept β so the plan shifts towards gradual, rewarding exposure rather than expecting immediate confidence.
No. Training works through consequences, and the practical version of leadership is controlling access to what the puppy wants β food, doors, attention, play β and being consistent about it. Confrontation produces a dog that complies while watched and stops when it is not.
Training a puppy is mostly a discipline about timing and difficulty, and both are things the handler controls. The puppy is doing exactly what the last half-second taught it. PawAtlas treats it the way it treats the behaviour problems in this series: the animal is not defying anyone, it is responding to the consequences it actually received rather than the ones intended, and the fix is to make the intended consequence the one that arrives.