The week-by-week desensitization protocol that treats separation anxiety in puppies — how to find the threshold, build duration, and handle regressions.
There is exactly one approach with evidence behind it: systematic desensitization — practising departures short enough that the puppy never actually panics, then extending them by seconds. Everything else, from leaving the TV on to "letting them cry it out", either does nothing or actively makes it worse.
The whole method reduces to one rule: never let the puppy rehearse the panic. A departure that ends in genuine panic sets you back further than three good sessions move you forward.
Before starting, make sure this is the right problem. Does my puppy have separation anxiety? separates true SA from normal distress, boredom and a full bladder — the protocol below is overkill for the other three.
Everything depends on one number: how long can your puppy be alone before distress starts?
Set up a phone recording. Leave. Watch it back and note the moment the body changes — pacing, whining, staring at the door. That moment is your threshold. For a puppy with real SA it is often 20–45 seconds. It can be less. That is fine, and it is not a verdict on your dog.
Work at 80% of the threshold, not at it. If distress starts at 40 seconds, practise 30-second departures.
| Week | Target | Sessions/day | What you are doing |
|---|---|---|---|
| 1 | 30s → 1 min | 5–8 | Short exits. Return before any distress, every time. |
| 2 | 1 → 3 min | 5–6 | Add 15–30s only after three clean sessions. |
| 3 | 3 → 8 min | 4–5 | Start varying which door and which time of day. |
| 4 | 8 → 20 min | 3–4 | Introduce the real getting-ready routine. |
| 5 | 20 → 45 min | 3 | Actually leave the building. |
| 6 | 45 min → 90 min | 2–3 | Ordinary absences. Keep logging. |
These are targets, not a schedule to force. If week 3 takes three weeks, it takes three weeks. The dogs that finish fastest are the ones whose owners went slowest early.
1. Return only during calm. Never open the door mid-cry. Wait for a three-second gap and use that. This is the single most-broken rule and the most consequential.
2. Decouple the departure cues. Pick up your keys and sit down. Put your coat on and make tea. Twenty times a day, going nowhere. Within a week the keys stop meaning anything, and you have removed the thing that starts the panic before you have even left.
3. One reward that only exists when you leave. A frozen stuffed Kong appears as you go and is taken away when you return. Alone time becomes the thing that predicts the good stuff.
4. Zero drama at both ends. No lengthy goodbyes, no ecstatic reunions. Boring departures, boring returns.
5. No above-threshold absences during treatment. This is the hard one. If you must be out for four hours in week 2, use a sitter, a neighbour, or daycare. One four-hour panic can cost you a month.
It will. Regression is part of the normal shape of this, not evidence of failure. The usual causes:
The response is always the same: go back a step, get three clean sessions, move forward again.
Now, if there is any self-injury — broken nails, bloodied gums, worn paws. Otherwise, if six weeks of consistent, correctly-executed practice has produced no measurable change.
For established cases, a veterinary behaviourist may discuss medication alongside the behaviour work. That is not a shortcut or an admission of defeat: for some dogs the anxiety is high enough that they cannot learn anything at all until it comes down, and medication is what makes the training possible.
Puppy AI does not attempt to treat this class of problem. Its escalation rules route separation anxiety, resource guarding, fear and aggression to a credentialed human by design.
Realistic timelines are in how long puppy separation anxiety lasts, and the causes, research and crate-training questions are covered in the main guide to separation anxiety in puppies.
Three fields per session: duration, whether the return was during calm, and the settle time. That is enough to answer the only two questions that matter — am I moving too fast? and is this working at all? — neither of which is answerable from memory.
It is also exactly the record a trainer or veterinary behaviourist will ask for at the first appointment, and turning up with six weeks of real data instead of an anecdote changes what you get out of that hour.