Most of what’s written about separation anxiety in dogs is advice. Informed advice, but advice all the same. What we have is something rarer: real session-by-session tracking data, drawn from cases we’ve worked on in Singapore over the past two and a half years. Every session logged, every duration recorded, every setback noted alongside what caused it.
This isn’t a survey. Nobody’s recalling how their dog did from memory months later. It’s a running record, built one session at a time, by the owners actually doing the work with their own dogs.
Here’s what that data actually shows: the real milestones, the real setbacks, and the parts of the process that don’t show up in a duration number alone.
How We Track Every Session
Every case we take on gets logged the same way. Owners run each session themselves, usually in short, structured stretches a few times a week, and record how long their dog managed alone, what the environment was like that day, and how the session went overall: a simple Great, Good, or Fair rating, with Fair reserved for a session where something genuinely didn’t go well, not just a slightly lower score. Across every session we track, the large majority land on Good or Great, about 84% combined, so Fair is very much the exception. Our Certified Separation Anxiety Trainers (CSAT) design the programme and guide it from session to session, but it’s the owner doing the actual work with their dog and logging what happened.
That distinction matters for how to read everything that follows: this reflects what owners observed and recorded about their own dogs, session by session, not an outside clinical assessment.
The training itself follows a structured, gradual approach: departure signals get reintroduced into a dog’s routine deliberately, one at a time, not all at once. We won’t walk through the exact sequence here. It’s easy to get wrong without guidance, and applying it incorrectly can set a dog back further than doing nothing at all. If you want the full method, that’s covered in our full desensitisation protocol.
Which signals show up varies household to household, but a few are common enough across the cases we track to be worth naming, several of them specific to how people actually live in Singapore. We go deeper on those in our breakdown of Singapore-specific separation anxiety triggers.
The metal gate (a second barrier most homes outside Singapore simply don’t have) is the single most common signal in our data. Electronic door access (a digital lock, tap card, or remote) comes next; for most dogs this is just how the door normally opens rather than something anyone’s identified as a specific trigger, though it can become one.
How Progress Typically Unfolds Within a Training Window
Every dog’s programme runs on its own timeline. In the cases we track, training typically spans three to seven weeks of near-daily sessions, with the first formal progress check happening around session six on average.
Within that window, here’s what the data shows: of dogs who cross the five-minute mark, 55-64% go on to reach fifteen or twenty minutes. Cross fifteen minutes, though, and that likelihood jumps sharply: around 86% of dogs who get past fifteen minutes go on to reach a full thirty. Fifteen minutes, not five, is the milestone that seems to matter most.
Every dog moves through these milestones at its own pace. Some settle into a shorter, steady duration that’s already a real, meaningful change in daily life, long enough to duck out for a coffee run without a worried dash back home. Others keep climbing well past the thirty-minute mark. Either way, the fifteen-minute threshold remains the clearest signal we’ve found that a dog is on track.
What Progress Looks Like Emotionally, Not Just in Minutes
Minutes tell only part of the story. The more telling changes show up in how a dog behaves during a session, and those tend to show up early, often before any real jump in duration.
At the very first session, most dogs show at least one classic stress signal: around six in ten camp out by the door, roughly four in ten bark when it closes, about a third pant or whine, and around a quarter freeze into a tense, watchful posture. Only about one in ten dogs starts out calm.
(Note: this is a ranked bar chart, not a pie/part-to-whole chart. Dogs often show more than one signal at once, so these percentages don’t sum to 100%. A pie would misrepresent that.)
By the third or fourth session, that picture looks very different. Barking drops to under 15% of dogs, panting to under 20%, and camping by the door falls below 30%. If barking or howling is the specific behaviour you’re dealing with right now, our guide on managing excessive barking covers what to do in the meantime. By this point, roughly 85% of dogs have had at least one moment in the session with no visible stress reaction at all. Many of the same signs owners first noticed before starting training begin easing right around here.
It’s usually around this same point that an owner’s own session notes shift tone, too: comments like “getting the hang of it” or “starting to get the game” tend to show up around session three or four, across cases that otherwise have nothing in common. A dog settling into a lying position, ignoring the door entirely, chewing calmly on a toy between steps, or keeping its head down for the whole absence are the strongest signals we see that things are heading the right way, often well before the duration numbers catch up. In some of the calmer cases, a dog has fallen fully asleep mid-session, which is about as clear a signal of comfort as it gets.
This pattern, a sharp drop in visible stress within the first handful of sessions, holds up consistently across the cases we track, regardless of how severe a dog’s anxiety was at the start, whether they’re on medication, or how their household is set up. We treat it as an informal metric in its own right, tracked alongside duration, because behaviour tends to give us feedback faster than the clock does. In a week where the duration number barely moves, this is often the clearest early sign that a dog is still heading the right direction.
Regression Is Universal, and Recoverable
Almost every dog we’ve worked with regresses at some point during training: a session that goes worse than the ones before it. That’s not a sign something’s gone wrong. It’s part of the arc.
A regression almost always traces back to something specific: a real departure that day instead of a practice one, a thunderstorm, two people leaving instead of one, an unfamiliar visitor. Once the trigger passes, dogs typically bounce back within one or two sessions, not a restart from the beginning.
We checked five commonly-cited disruptors against the actual session ratings.
Domestic helper or visitor is a clear case: sessions where someone unexpectedly appears, or walks past the house, tend to trigger the dog, cutting short the stretch of calm before it reacts. A session like that can still get rated Good or Great overall: CSAT specialists coach owners to rate around the core separation-anxiety signals, not a one-off reaction to someone else nearby, so a shorter calm stretch doesn’t automatically pull the rating down. Sessions with someone else unexpectedly appearing still ran a Fair rate several times higher than sessions where it was just the dog and the owner. Vet visit points the same direction: dogs that visited the vet earlier that day were more likely to react, with a shorter stretch of calm during that session.
Thunder or rain and construction noise are a different kind of confirmed. The session-rating check alone doesn’t show a clear effect for either, but our Certified Separation Anxiety Trainer’s clinical judgement is that both are genuine disruptor mechanisms: thunder phobia is a well-documented reaction pattern in dogs, and noise sensitivity (construction work, delivery visits, and similar sudden sounds) can push a dog past its threshold, the point where it stops coping and reacts, within what’s otherwise a good session. These disruptors are usually obvious when they happen, so the session still tends to get rated Fair, Good, or Great overall: everyone involved can see the dog was calm right up until the disruptor occurred, and rates the session on that basis.
Daycare day is the one that didn’t hold up on either front. No mechanism was offered for why it would disrupt training, and the data points the other way: dogs settle in just as well, sometimes better, after a daycare day. Here’s that check specifically, alongside the two categories the rating data does confirm directly.
The bigger point isn’t any one of these results. It’s that a category sounding like a plausible disruptor in the qualitative notes doesn’t mean it shows up in what actually happened in the room.
If a session goes badly, the useful question isn’t “did we lose all our progress.” It’s “what happened that day.” Nine times out of ten, there’s a clear answer, and the session after looks like the ones before the setback.
Does Medication Help? What We’ve Actually Observed
Medication has a real place in separation anxiety training, particularly for more severe cases: it can help a dog stay calmer and further below their anxiety threshold, which makes the desensitisation process safer to run.
What it doesn’t appear to do, based on the cases we track, is guarantee a faster or smoother path. Medicated dogs in our data show no single consistent pattern: some progress quickly, others plateau or regress despite the medication, and some of our more heavily medicated cases saw a higher rate of difficult sessions along the way, which tracks with medication typically being reserved for the harder cases to begin with.
The honest takeaway: treat medication as a tool for managing severity, not a predictor of how training will go. Any decision to use it should sit with your vet or a veterinary behaviourist, alongside the training itself.
What’s Next: Expanding This Research
Two things we’re actively working on.
First, we’re reaching out to past clients to learn how their dogs are actually doing months after their programme wrapped up: real-world, fully unsupervised conditions, not a training session. That’s the piece of the puzzle we don’t have yet, and we plan to close that gap.
Second, we’re starting to record breed information for new clients going forward, to see whether any Singapore-specific patterns emerge over time.
Neither of these exists yet, but they will. We’ll publish updated findings as this research continues.
If you’re noticing any of these patterns in your own dog, book a separation anxiety assessment with one of our CSAT trainers. The earlier a programme starts, the more of this process you get to see for yourself.
