Why does my dog obey at home but not outside?

Stormie's file is where the concept is named out loud to a family, and where the trainer says he does not like the name. The three Ds - distance, duration, distraction - with the rule attached: you do not add the second until the first one is good, and duration is where they start. She holds place for about ten minutes, which he calls a goal time, and he says that now is when the distractions come in. An earlier part of the same lesson gives the order concretely: place work first because impulse control is her weak point, then distance, then more people, then walking past dogs. And her file is one of the two that force a fourth D into the list. On a hot day she downs on grass in seconds and refuses the concrete slab; the handler says outright that she is cool with the grass and does not want it on the concrete, works reps on the slab rather than around it, and two days later states it as policy - the commands have to be learned on concrete and then everywhere else. Her file also supplies the only deliberate use of the dog's own state as a variable: she is tired, so he walks her past people specifically to find out whether the tiredness helps the impulse control or breaks it.

Luna is where the surface stops being an observation and becomes an experiment. Her down is fluent; in a new place, on concrete she has never been asked to lie on, she fights it. The trainer says he has never done it on concrete before, notes that she did a couple of quick downs in her session the other day and calls the difference very weird, and then moves her onto grass to find out whether the ground is what she is arguing with. Her first owner lesson had already stated the principle - start on grass, because dogs sometimes will not down unless they are on grass until they really know the command, then move to concrete - and a third session confirms it in the plainest possible terms, that she prefers the grass, which makes sense, because it is hot. The same lesson carries the companion refusal, which is really a statement about hidden variables: he will not say sit before the down, even though it would put her in an easier position, because she would learn that the down needs a sit first.

Louie’s surface example begins with the trainer explaining that some dogs need to practice on grass first. In a later shared session with Sage, the trainer discusses Louie responding more quickly after moving from blacktop to grass. At 11:25, the discussion turns to Sage and the trainer reports having tested the ground’s temperature. That report supports checking conditions before interpreting a response; it is not a measurement of surface temperature or independent visual verification of how the ground was checked. Another Louie session contrasts his heel when traveling somewhere with practice in the training yard, which the trainer identifies as an area to keep proofing.

Sage's set gives the heel ladder as a plan - a little longer, then abrupt stops, then fast and slow - and makes the sequencing explicit: another session first, and only if she is doing well each time do the criteria go up. Her intake description names which two of the three she was missing: no command at verbal-only fluency, no duration, and off the command at any distraction. And the joint session with Louie contains the cleanest diagnosis of a hidden variable anywhere in this corpus. A down is slow; the handler wonders whether it is the blacktop; he moves the dog to the grass; it gets quicker; he concludes that the surface was a big part of the problem. That is a hypothesis, a test and a revised conclusion inside ninety seconds, said out loud, about something that would ordinarily be filed as stubbornness.

Dolce's file adds gradient, and the way it is added is the most honest moment in this concept. Mid-session, off leash, on uneven ground, the handler asks for level ground and then says he had not thought about that - proofing for downhill and uphill, or uneven ground, because it can definitely change the way a dog does a heel. A variable discovered live, named, and folded into the list by somebody who could simply have moved on. Her owner lesson gives the standard version alongside it: the heel is made longer and longer in different areas, so that she learns that no matter what is going on she has to hold it for however long she is asked. And a later session raises distance to its limit by the handler leaving the scene entirely, with the e-collar kept on as the only thing that can reach that far.

Dutchess is the case that separates two things the standard list collapses. Distance, in the proofing sense, is a difficulty dial. Proximity, for her, is a relationship: she wants to be touching her handler, so working three feet away is genuinely hard, and the trainer's reading is that she is trying - she has to learn she cannot be that close. He gets there by elimination rather than assumption, and says so: she is fine with duration if he is close to her, for everything, so it really is the distance part. The rest of her file is textbook - about twenty seconds of held down before stepping away, one criterion raised at a time, distraction added to a behaviour before duration and distance are asked for at all - and her second lesson raises distraction in one deliberate step from two easy dogs to the whole household at once, in the room where the problems live.

Kodak supplies two things at opposite ends of the concept. The first is the clean statement of the standard model, said on a walk: he is starting to add the three Ds - distance, duration and distraction - to the down. The second is the case none of the three explains. Off a cot, in a different room, the sit is simply gone: he knows how to sit, he does it on the cot with no problem, and anywhere else he is different. No distance, no duration, nothing added - only context, which is not on the list. His pack-walk file then proposes the axis that matters most for social work: difficulty is raised along the amount of escape available, so outdoors comes first because either dog can get away, and indoors comes last because a room removes the retreat and adds everybody else's tension. And on the heel he is worked on the specific turn where he is on the outside and has to travel further, which is a distance variable hiding inside a heel.

Bullet's file has the clearest account of how a criterion gets raised, and it puts the reset in the middle of the mechanism rather than beside it: the dog gets enough time to come completely back to baseline - wander, get pets, sniff everything - and only then does the criterion go up, so that he thinks he has figured it out each time. Duration or distraction, one of them, and over time you have a heel that is more and more solid. His file also contains a variable that is not distraction: pace. He knows the heel, and if the handler speeds up he falls behind - not forging, not crowding, lagging - which is a fault only a change of speed produces. And a place hold with a cat brought up to the cot is a rare instance of the size of a step being named out loud as bigger than the last one.

Bo shows the dial being turned down, which is the half most owners never see. On a public trail his whole criterion for success is that he stops when the handler stops - one element, named in advance, with the distraction left high. In a busy store with two strange dogs and machinery, the trainer decides he has pitched the criteria too high, says so, gives the dog a break, and then gets the heel by lowering the criteria a little - saying that out loud as he does it. And in a crowded shopping area the environment is named as the cause of the failures rather than the dog: he is very distracted here, and the criterion shrinks to one good turn.

King's proofing is done with the thing he actually wants, which is what makes it count. The ball goes in a pocket, then it is thrown, then it is thrown while he is on place, then the handler runs at it himself - a distraction raised in graded steps inside one exercise, with the dog holding. The trainer's summary of a long place session names two of the three Ds together and is careful about the third variable in it: he held for a longer duration with more distractions, there were dogs walking by, a little far away, but still good to see. And the philosophy behind it is in one line - that is exactly what he wants to see, the temptation part, you need temptations.

Zeus's file states the heel as a three-stage model rather than a list of dials, and the third stage is this concept exactly: once he knows the position and knows it is his job to keep it while it moves, you try to shake him out of it, and you give consequences either way when he keeps it. The remaining work on the heel is then named as two things and no more - reinforce it consistently, and proof it under distraction. His first owner lesson gives the venue ladder that goes with it: the facility's training room and yard first, then trails and parks where he will see other dogs but there is room, and stores towards the end of the programme.

Finn is the dog for whom the order is reversed and the handler says so. With most dogs you can add distance and distraction fairly quickly; with him it is duration and that is it, close up, and distance goes on afterwards. The reason is in another session: distance is the criterion that breaks his sit, because stepping away is its own problem for some dogs, so the criterion is lowered back to sitting next to him. And his file contains the plainest statement in this concept of what a change of context actually costs: everything they had to go through to get basic things indoors has to be gone through all over again outdoors, and the trainer describes it as a reset rather than as a regression.

Deuce's file adds the variable an owner will never think of: another person the dog loves. His recall is nice around the trainer, and once his owner entered the picture his control went down a lot - the presence of a particular human treated as a difficulty dial in its own right. He also shows a failure being converted into a proofing plan rather than a correction: a stranger actively engaged with him, so the blown recall is not punished, but it is named as something that needs more proofing. And tiredness appears again as a condition rather than an excuse - the point being that the rules should still be clear to him when he is tired.

Lola's case holds the three Ds against a nervous dog's ceiling. Distance is withheld because her nerves block it, and duration is raised instead - with the reason given in the negative: he does not want her to learn that coming out here to train is a miserable experience. Her go-home lesson turns the same idea into a three-phase ladder a family can run: inside behaviour first, then the back garden, then out and about, stores and hikes. And her second lesson adds the criterion that decides when a duration repetition is actually finished, which is not a clock at all - the anxious stare has to go, and until it does it is not really a smooth repetition of the command.

Roxy's sessions treat environment and latency as the criteria being raised once the behaviour itself is done: she has learned it to a relatively good extent, so this is a different environment, proofed out in public, smoother and quicker on verbal only. Distraction is raised deliberately and announced - they are going to challenge her - while distance and duration are left alone, which is the one-at-a-time rule being followed in practice rather than described. And the household's other dog supplies the distraction for the sit, a few feet away and working, with a recall used as the reset.

What are the three Ds?

Distance, duration and distraction - how far away you are, how long you ask him to hold it, and what else is going on. A trainer names them to a family in this corpus and adds, in the same breath, that he does not love the name but it is what they call them. The rule that goes with them matters more than the list: you add one at a time, and you do not add the next until the first one is good. Duration is where he says they start. On place that means holding for longer before anybody steps away or anything interesting happens; ten minutes is described as a goal time. Only then do the distractions come in.

He lies down on grass in a second and will not do it on concrete. Is he being stubborn?

Probably not. This is the most common hidden variable in the whole corpus and it comes up on four different dogs. One trainer starts every down on grass as a matter of course, because dogs sometimes will not lie down on anything else until they really know the command, and moves to concrete later. Another says plainly that some dogs are weird about it and you have to do it in the grass first. A third, watching a slow down on blacktop, wonders whether the surface is the problem, moves the dog to grass, gets a quicker down, and concludes the blacktop was a big part of it. And a fourth checks the ground with his own hand before asking a dog to sit on it. So before you turn anything up, put your palm on the ground, and try the same repetition three feet away on grass.

Which one should I raise first?

Duration is the stated default, and the reason given is that it builds the impulse control the other two will need - five or ten minutes on place is described as a whole workout for a dog who is not naturally patient. But the trainers do not follow it blindly, and the exceptions are instructive. For a frightened dog the order was inverted: with most dogs you can add distance and distraction fairly quickly, with him it was duration only, close up, and distance afterwards. For another dog the diagnosis ran the other way - she was fine at any duration as long as the handler was near her, so distance was the only hole and the only thing worth working. The useful version of the rule is: raise the one this dog is worst at, one at a time, and know which one you are raising.

How do I actually raise one, in practice?

Get the repetition, then let the dog come all the way back down before you ask again. One trainer spells it out: the dog gets plenty of time to reset - walk around, follow him about, get some pets, sniff everything on a walk - and only after he is completely back to baseline does the criterion go up. The point of the reset is not kindness, it is that the dog gets to feel as though he has figured it out each time, and that is what makes the next raise survivable. Then you raise one thing - a bit longer, or a bit more going on - and over time you have a command that is more and more solid. If it fails, come down rather than pushing: one trainer in a busy store says out loud that he pitched the criteria too high, gives the dog a break, lowers them a little, and gets the heel immediately.

What else changes a command besides those three?

More than most lists admit, and this corpus keeps finding them one at a time. The ground he has to lie on. The slope - uphill, downhill and uneven ground change the mechanics of a heel, and the trainer who noticed it said he had not thought of it before. Your walking speed - one dog holds a slow heel perfectly and lags on a fast one, which is a fault only the pace produces. How much room there is to get away, which is why two dogs meeting outdoors is easier than two dogs meeting in a living room. How close you are standing, which is not the same thing as distance-as-proofing: one dog in this record wants to be touching her handler, so three feet away is genuinely hard for her. And the dog's own state - one handler walked a tired dog past people specifically to find out whether the tiredness improved her impulse control or destroyed it.

He does it perfectly on his bed and nowhere else. Which D is that?

None of them, and that is the point. Two handlers in this corpus stand a metre apart and demonstrate it: he knows how to sit, he does it on the cot with no problem, and off the cot he will not - bring him anywhere else and he is a different dog. No distance was added, no duration, no distraction. Only the place changed. That is a context failure rather than a criterion failure, and the fix is not to raise anything but to run the same teaching again in the new place. A nervous dog's file puts it in the same terms: everything they had to go through to get basic commands indoors has to be gone through again outdoors, and the trainer calls it a reset rather than a regression.

What should I use as a distraction?

Whatever your dog actually wants, and preferably something already in your life rather than something you set up. In this corpus the distractions that count are a ball thrown past a dog on place, then the handler running at it; other dogs walking by at a distance he can just about tolerate; the front door opening; a live cat brought up to a place cot; a person who genuinely engages with the dog. One trainer sums up the standard in one line - that is exactly what he wants to see, the temptation part, you need temptations. And one warns against the cheap version: a repetition against something the dog is only mildly curious about proves nothing, because curious is not the same as pumped.

Raising two things at once - stepping further away and asking for longer, or adding distance in a busier place - and then reading the failure as disobedience.

Treating a refusal to lie down on concrete, tile or hot tarmac as stubbornness, when the dog has never been asked to do it on that surface or the surface is too hot.

Asking for the next repetition before the dog has come back to baseline, which removes the sense of having figured it out that the whole escalation depends on.

Helping the dog into position with an extra cue - saying sit before down - which teaches him the command needs the other command first.

Introducing two dogs to each other indoors, which removes the retreat and raises the difficulty at exactly the moment everybody thinks they are being careful.

Correcting a dog who wants to be close for being close, when proximity is the thing that has to be trained rather than the thing being refused.

Forgetting that the people the dog loves are a distraction. A recall that holds for a trainer can collapse the moment a family member walks into the picture.

He downs instantly on the lawn and will not do it on the patio, the pavement or the kitchen floor.

The command holds while you are standing next to him and falls apart the moment you walk away.

His heel is fine at your normal pace and he lags the moment you speed up.

He is reliable for one person in the household and not for another, or reliable until a particular person walks in.

He holds a command in one specific spot - his bed, his cot, the hallway - and nowhere else at all.

Your two dogs are fine in the garden and tense in the living room.

He will only down if you ask for a sit first, and you did not mean to teach him that.