Reading Separation Anxiety on Video: What to Look For

6 May 2026 · Rachel Trafford

Reading Separation Anxiety on Video: What to Look For

Separation related behaviour is one of the cases best suited to video, for a simple reason. By definition it happens when the owner is not there. The behaviour you most need to see is the behaviour nobody is present to observe. A camera is often the only honest witness.

And here is the thing that has changed, quietly but completely, in the last few years. Cameras are everywhere now. Ring doorbells, nanny cams, indoor pet cameras, old phones propped on a shelf. Three or four years ago this kind of footage was rare and you often had to ask an owner to set something up. Today most owners arrive already holding plenty of it. They have clips, they have neighbour complaints about the barking, they have the whole sorry record of what separation distress does to a household. Our job is not to manufacture more of it. Our job is to use what already exists, and to use it well.

A note on terminology before we go further. Separation anxiety is used loosely in everyday language, but separation related behaviours have several different underlying causes, and not all of them are anxiety. Reading the video well means resisting the urge to label too early, and thinking instead about what is driving the behaviour and what function it serves for the dog.

A welfare line that comes first, do not gather data by causing distress

This matters more than anything else in the post, so it goes near the top. We do not put a dog into distress in order to get good footage. We do not flood. We do not ask an owner to leave the dog for longer than it can cope with so we can capture a dramatic clip for the file.

The whole philosophy of good separation work is keeping the dog under threshold, below the point where it tips into panic, so that it can actually learn to feel safe. Deliberately exceeding that threshold to gather information would undo the very thing we are trying to build, and it would harm the dog. We do not do it.

The good news is that we almost never need to. Most owners already have more footage than we can use, gathered before they ever found us, back when nobody knew what was happening and the cameras were just running. That existing data is where we work from. From the moment a dog is in our care, the aim is to gather information without ever pushing the dog over the edge to get it.

Start before the departure, not after

The most common mistake in assessing separation footage is starting the clock at the moment the door closes. The most informative period is often the ten to fifteen minutes before the owner leaves.

Watch for anticipatory behaviour during the departure routine, picking up keys, putting on shoes, the bag by the door. A dog who is already pacing, panting, or shadowing the owner closely during these pre departure cues is telling you that the distress is conditioned to the signals of leaving, not just to the absence itself. This shapes the whole plan, so it is worth the footage starting early. Owners with a doorbell camera or a hallway camera very often have this part already.

The first few minutes after departure are diagnostic

The period immediately after the door closes is where the most useful information sits. Look for the onset and the trajectory.

How quickly does distress appear? Immediate onset, within seconds to a couple of minutes, suggests a strong conditioned response. A gradual build over many minutes suggests something different.

What is the first behaviour? Vocalisation, pacing, scratching at the exit point, attempting to follow, settling then re rousing. The sequence matters.

Is there an exit focus? Dogs with separation distress often orient strongly toward the point of departure, the door the owner left through, a window, a gate. Destructive behaviour concentrated at exits is a different picture from generalised destruction.

The signals worth pausing on, including the quiet ones

When reviewing separation footage, these are the signals worth slowing down for.

Panting with no thermal explanation, a key physiological marker of arousal. Salivation, drool pooling, wet patches on bedding, a wet chin. Lip licking and yawning out of context, the low level stress signals that often precede the more obvious behaviours. Pacing, fixed routes, repetitive circuits, an inability to commit to lying down. Whale eye and facial tension during the settle attempts. A lowered, tense, ready to move body rather than a loose settled one. And repeated re rousing, the dog lies down, then gets up again, over and over, never achieving genuine rest.

There is a set of signals that is easy to miss because it is defined by absence rather than action, and it is worth its own mention. A dog whose system does not feel safe will often not eat, will not take a chew, will leave a favourite treat untouched, and in some cases will not even drink water while alone. Refusing food, chews, and water is not fussiness, it is a marker of a dog too anxious to do the ordinary things a settled dog does. If an owner says the stuffed Kong comes back full every time, or the water bowl is untouched after four hours alone, take that seriously. The dog is telling you it did not feel safe enough to relax.

Distinguish distress from boredom, and anxiety from other causes

Several quite different situations can look superficially similar on first viewing.

Distress versus under stimulation. A genuinely distressed dog typically shows physiological signs, panting that is not heat related, salivation, trembling, an inability to settle, repetitive behaviour, and the refusal to eat or drink described above. A bored or under exercised dog destroying things is often calmer in body, more exploratory, and will frequently settle once the interesting thing is destroyed. The body tells you which you are looking at.

Panic versus frustration. Some dogs are not anxious about being alone in general but are specifically frustrated at being barred from the owner. The behaviour can look identical at first, but the frustrated dog's focus is on the barrier and the owner, and the behaviour often eases if the dog can see or reach the owner, whereas true separation distress persists.

Single event versus sustained. Watch the whole footage, not just the first burst. Many dogs have an initial period of distress that resolves into settling within twenty to thirty minutes. A dog who settles is a very different prognosis from a dog who sustains or escalates distress for the entire absence.

The real value, comparing departures over time

Here is where separation work becomes something more than a single assessment. A clip on one day is a snapshot. The thing that actually drives a case forward is comparing departures over time, the same dog, the same kind of absence, watched week after week, so you can see whether the picture is genuinely improving.

Is the onset of distress getting later? Is the peak less severe? Is the dog settling sooner, eating the chew it used to refuse, drinking the water it used to leave, achieving real rest where before it only paced? These comparisons are the spine of a separation programme. They tell you whether the plan is working, they protect against false optimism, and crucially they give the owner something concrete to hold onto during a long and tiring process.

Why most separation cases fail, and what actually carries them

It is worth being honest about why separation cases go wrong, because it is almost never the dog.

Most failed separation cases fail for two linked reasons. The first is the absence of a real plan, a structured, staged progression that keeps the dog under threshold throughout. The second, which follows from the first, is the owner progressing too soon. Without a clear plan and clear stage markers, an owner naturally pushes on, leaves the dog a little longer because last time seemed fine, and the dog tips over threshold during the very behaviour modification that is supposed to be protecting it. Going over threshold during the mod is not progress, it is a setback, and it is the single most common way these cases unravel.

Good separation work does the opposite. It plans the stages carefully, it protects the dog from distress exposure rather than testing how much it can take, and it only moves forward when the data says the current stage is genuinely solid.

But here is the part the textbooks underplay. Separation work is hard, slow, demoralising work for the owner. Progress is measured in seconds and minutes, the gains are invisible day to day, and the temptation to rush or to give up is constant. What carries a separation case is not just the protocol, it is the relationship around it. Close working. Frequent feedback. Clear stage markers so the owner knows exactly where they are. And mini celebrations at every genuine milestone, the first time the dog settled, the first chew finished, the first ten clean minutes. Owners doing this work need cheerleading and steady hand holding, and providing that is exhausting for trainers. It is one of the real pain points of taking on separation cases, and any system that lightens that load, that documents the stages and surfaces the wins so they can be celebrated, is doing something genuinely useful.

Where AI video analysis fits, and where it does not

This is exactly the kind of assessment where a structured, frame by frame analysis can help a trainer, by surfacing the small signals, the lip lick at 0:40, the panting onset at 1:15, the untouched chew, that are easy to miss on a single watch, and by documenting the trajectory the same way every time so departures can be compared across weeks.

But it is also exactly the kind of assessment where the model's limits show. An AI can note that a dog is panting and pacing and has not touched its water. It cannot know the dog has no pain history, that the owner's departure routine changed last week, that this is day three of a careful plan, or whether this clip represents improvement or regression compared to the last one. The distinction between panic and frustration, between distress and boredom, often turns on context the camera cannot supply, and the decision about when it is safe to move to the next stage is a clinical judgement that belongs firmly with the trainer.

So the right use is the same as everywhere else in this work. Let the analysis do the thorough, consistent first pass and the patient comparison across departures. Let the trainer bring the judgement, the history, the plan, and the human cheerleading that actually gets an owner and a dog through to the other side.

MyCanine360 includes a dedicated separation related behaviour analysis mode, built to work from the footage owners already have, under threshold, with consistent assessment across departures so progress can be tracked and celebrated. The trainer always reviews and edits the assessment before it informs a plan. [Learn more / try it].

Related reading:

• The Body Language of a Dog About to Bite: Pre-Conflict Signals on Video
• Why "My Dog Seems Fine Now" Is the Trainer's Hardest Problem