Your dog nails sit-stay every single time in the kitchen. Then you get to the dog park, a stranger waves hello, a squirrel pops out of a rabbit hole, and suddenly your dog acts like you've never asked "sit" in their life...in fact, who even are you compared to this squirrel?
This is a proofing problem.
Proofing is the process of building reliable behavior across different contexts, environments, distractions, and people. Proofing is the bulk of what most dog training should be. Most owners skip it because they don't realize how important it is and because their mental model of training is sessions at home.
The response reliability gap
Ian Dunbar has a simple way to measure where your dog's training actually stands. Take any single command (sit, down, stand) and give it ten times in a row. Count how many responses you get, how many times your dog follows the command. Divide responses by commands. That's your response reliability rate.
Most dogs and owners are shocked by the number. The dog that feels "100% reliable" is often running at 60–70% when tested honestly. Even top competition dogs rarely hit 100%.
Most training happens in the same spot, in the same order, with the same tone of voice, which means the dog isn't listening to the word; they're reading the pattern — a dog who's learned to guess well in familiar conditions, not one who actually knows the command.
Your dog is predicting you, not listening to you
Dunbar makes a point that cuts to the heart of the problem: training is all about anticipation, and dogs can anticipate the bad as well as the good.
If you cycle through sit and down in the same order over and over (what Dunbar calls "dog push-ups"), the dog stops responding to the word and starts predicting the sequence. Ask them to sit and they might lie down, because in their experience, down almost always follows sit. They got the pattern right, even though they got the command wrong.
The fix is randomization, and it requires teaching a third behavior. Dunbar's recommendation: work sit, down, and stand together, randomized. If you only have two behaviors in rotation, the dog has a 50% chance of being right just by alternating. Add stand, and now they actually have to listen. They can't game it.
This is also why teaching "stand" gets pushed to the back burner in most basic training, and recall and down reliability suffer for it. You need at least three behaviors in the pool to break pattern-prediction and build real word comprehension.
Proofing in the real world: make it gradual and use distance
Once a behavior is solid at home, the natural instinct is to take it to a busy environment and test it. But that's usually too big of a jump.
Michael Ellis describes a cleaner approach: work at a greater distance from anything you can't control. In a class setting with a friend's dog, you have control: you know the dog, the handler, and the level of unpredictability. In the world, you don't. A stranger's dog might lunge. A child might run over. Those uncontrolled variables will overwhelm a dog whose behavior hasn't been proofed at lower levels of distraction yet. So as soon as your dog sees that other dog 50 yards away, ask your dog to sit, then gradually build closer. This could be in the same session, but it may take multiple sessions. Even if you can get closer and still get a sit in the same session, you'll want to do this many times to really proof the sit when another dog is around.
I would also add to gradually expand the new environments. Proof in all the rooms in your house before proofing outside. Proof in the back yard before the front, and the front yard before proofing around the block. And just try to be in tune with how quickly or slowly your dog is catching on.
The two ways proofing fails
Ellis identifies two common failure modes when people try to proof young dogs around distractions.
The first: the dog has a bad experience, gets overwhelmed, rushed by another dog, or scared, and now the environment itself is a source of stress. You've proofed in the wrong direction.
The second is subtler. You let the dog meet every dog, greet every person, explore every distraction freely. Now the dog wants to drag you to every dog it sees. The world becomes more interesting than you are, which tanks your obedience and your recall.
Both failure modes come down to what the dog was allowed to practice.
Turn distractions into rewards
The shift that makes proofing work: stop treating distractions as obstacles to training, and start using them as the reward.
Dunbar demonstrates this clearly in his puppy classes. In the middle of a group play session, he calls a dog, has them sit, then says "go play." The distraction (the other dogs, the chaos) becomes the payoff for compliance. The dog learns that coming when called doesn't end the fun. It's a brief check-in, and then the play continues.
This works with any high-value distraction: a favorite toy, sniffing, another dog, running. The pattern is: cue the behavior, reward with the distraction. Call the dog away from the smell, let them sit, then release them back to sniff. What they want becomes what they earn.
Boundary proofing: practical safety applications
One underused form of proofing is boundary work: teaching a dog that certain thresholds (the front door, the end of the driveway, the car door) require a sit before crossing.
Dunbar's method: every time you approach the boundary, ask for a sit. Practice it repeatedly until the sit before the threshold is automatic. Then comes the proofing step: you walk to the boundary without stopping and watch what the dog does.
If they've been trained and proofed correctly, they sit without being asked. The boundary itself becomes the cue — the kind of reliability that holds in an emergency, when you didn't remember to ask for it.
Socialization as proofing: it doesn't end
Ellis makes a point that often gets overlooked: proofing is not a puppy phase. It's an ongoing process.
Dogs that are well socialized as puppies can regress if owners stop exposing them to new environments. A dog raised in the city, moved to the country, and brought back to the city years later may suddenly be overwhelmed by environments they once navigated easily.
The environments your dog needs to be reliable in are the environments you need to keep exposing them to. Yes a dog does start to contextualize; a supposed rule of thumb is that if your dog reliably learns a skill in 7 differentiated environments, then it will be able to do that skill in most environments.
But you never know when you might meet a new distraction, and if you don't continue to proof, the reliability of your dog's skills will dwindle. Make it an ongoing, easy activity that's integrated into your flow with your dog.
What proofed actually looks like
The end goal Dunbar describes: the response becomes the reward. When sit has been trained well enough, rewarded in enough contexts, paired with enough good things, proofed in enough environments, the dog sits because sitting itself has value. It precedes going for a walk, getting a treat, being released to play, being let off leash. The behavior becomes self-reinforcing, and you stop needing food in your hand to get compliance.
Getting there takes intentional and consistent practice, but it's actually pretty simple, something you can take a lot of pride in, and completely worth it.
Doggo Discs are built for this kind of work, easy to carry, quick to deliver, and high enough value to compete with real-world distractions when you need them most.
