Not All Cocaine Smells the Same. And That’s the Problem

A real training scenario that raises an uncomfortable question about what our dogs actually recognise

I always learn from seminars — even when I’m the one delivering them.

I say it often in the detection dog world: the day you think you know it all is the day you should step away. No one knows it all. Every day is a learning day. And every now and then, something happens that leaves you asking questions you don’t immediately have the answer to.

This was one of those moments.

I’ve just delivered a three-day Detection Defined™ seminar. It was a busy one — ten dogs in total, which is more than I would usually like to work with — but it did give me the opportunity to see a wide mix of teams. There were nosework and sport dogs, alongside law enforcement teams from police, customs and the prison service. A real cross-section of the detection world, all in one place.

It was during an exercise with a Customs drug detection dog that things started to become unclear. Not just for me, but for the handler, and for the other experienced observers watching.

Exercise One

Two samples of real heroin, provided by Police:

• 10g heroin placed in the bottom of a window blind
• 10g heroin placed behind a framed picture (approx. 1.2m height)
• Time placed: 10:13 hrs
• Search commenced: 12:04 hrs
• Soak time: 1 hour 51 minutes
• Location: Large music classroom in a primary school

When the dog began searching, there was nothing about the scenario that suggested difficulty. The quantities were significant, the placements were straightforward, and the soak time was more than sufficient for odour to develop.

The dog searched for over twenty minutes and passed close to both hides on multiple occasions. But there was no change of behaviour. No shift in breathing, no head movement, no focus, no indication. Nothing at all to suggest the dog had encountered target odour.

It was confusing. For both of us.

Everything about the setup suggested this should have been a routine find, yet the dog worked as though there was nothing there.

Exercise Two — Same Dog

Later that afternoon, we ran a second scenario with the same dog:

• 10.45g cocaine (Police sample)
• Placed inside a plastic pen box on a desk
• Box not sealed
• Time placed: 15:09 hrs
• Search commenced: 16:17 hrs
• Soak time: 1 hour 8 minutes
• Location: Classroom environment

The dog worked the room and passed the desk multiple times.

Then, on one occasion, it physically opened the pen box with its nose, placed its nose directly inside the box, and then simply withdrew and continued searching.

No change of behaviour.

No hesitation.

No recognition.

No indication.

At that point, the question changed.

It was no longer “why didn’t the dog find it?”

It became: how does a trained drug detection dog place its nose directly on source and fail to recognise it?

Exercise Three — Same Dog

The following exercise added another layer to the situation:

• 10g cocaine (Customs training sample — the same type used routinely by this team)
• Placed in the pocket of a pair of dungarees
• Pocket zipped closed
• Dungarees hanging on a peg amongst multiple other garments
• Time placed: 09:10 hrs
• Search commenced: 10:45 hrs
• Soak time: 1 hour 35 minutes
• Location: Basement changing room area

This time, the environment was arguably more complex. Multiple garments, competing odours, and a concealed, zipped hide.

And yet within 30 seconds of entering the search area, everything changed.

The dog showed a clear change of behaviour, immediately began working the odour, and quickly sourced it. A clean, confident freeze indication followed.

No hesitation. No confusion. Just recognition and clarity.

Exercise Four — Controlled Comparison

To explore this further, I set up a more controlled test.

The aim was to remove as many variables as possible and present the dog with two near-identical scenarios, differing only in the sample.

• Locker 44:
• 10g methamphetamine (Police sample)
• Placed in a black rucksack
• Located in the top front zipper pocket
• Pocket zipped closed
• Rucksack placed in the bottom of the locker
• Locker door left slightly open
• Time placed: 14:03 hrs
• Locker 79:
• 10g methamphetamine (Customs sample)
• Placed in a black rucksack
• Located in the top front zipper pocket
• Pocket zipped closed
• Rucksack placed in the bottom of the locker
• Locker door left slightly open
• Time placed: 14:07 hrs

Both lockers were set up to be as close to identical as possible. Same type of bag, same pocket location, same concealment, same presentation. The only meaningful difference was the source of the sample.

The lockers were positioned at opposite ends of the room to avoid overlap.

At 15:38 hrs — 1 hour and 31 minutes later — the dog was deployed.

The dog worked the lockers methodically, screening along them.

It passed Locker 44 — the Police sample — multiple times.

No change of behaviour.

No interest.

No recognition.

But when it reached Locker 79 — the Customs sample — everything changed.

The dog showed clear recognition and gave an indication.

We can rule out airflow, access, and environmental difficulty. Across all four exercises, the odour was available. The dog worked the areas, passed the hides, and in one case made direct nose-to-source contact.

So we are left looking at something else.

Additional Observation

During the seminar, I also observed a police drugs dog working a separate scenario.

• 10g cannabis
• Placed inside a children’s school bag
• Bag positioned on a classroom desk
• Fully accessible from all sides
• Soak time: over one hour

Despite the simplicity of the setup, the dog failed to recognise the odour.

No clear change of behaviour.

No indication.

What makes this more notable is that this was reportedly a sample the dog had previously been trained on.

The exercise itself was intentionally simple. I had not seen the dog or handler work before, so this was designed as a straightforward assessment of both.

At a later point, I was accused of setting the dog up to fail—which is a difficult position to understand given the nature of the setup.

Importantly, a Customs drug dog handler from another country observed the exercise from start to finish and independently confirmed that the scenario was uncomplicated and should not have presented a problem for a trained police drugs dog.

This wasn’t part of the main sequence of exercises—but it adds another layer to the question.

Because it suggests that what we observed may not be isolated to one dog, one sample, or one scenario.

Generalisation

It would be easy to look at these exercises and put this down to generalisation.

And to a degree, that’s a fair consideration.

If a dog has only ever experienced odour in a particular context — certain packaging, certain environments, certain types of setup — then when that picture changes, recognition can break down. What we saw in the first two exercises could be interpreted as exactly that.

A dog encountering odour outside of its familiar picture and failing to recognise it.

But this is where it becomes important not to stop at the first explanation that sounds reasonable.

Because generalisation is not separate from training.

It is part of training.

A dog that truly understands odour should be able to recognise it across variation.

Different environments.

Different packaging.

Different presentations.

That is what generalisation is.

So if a dog fails to recognise odour when that picture changes, the question isn’t simply:

“Is this a generalisation issue?”

It becomes:

“Why hasn’t generalisation been achieved?”

And when we look at all four exercises together, the picture becomes clearer.

The dog did not struggle with the environment.

It did not struggle with access.

It did not struggle with complexity.

It struggled when the odour profile changed.

Which brings us to the real issue.

Sample Profile Difference

What these exercises show, very clearly, is the impact of sample profile.

The dog didn’t suddenly improve.

The environment didn’t suddenly become easier.

The task didn’t suddenly become clearer.

The only meaningful difference was the sample.

In the first two exercises, the dog was presented with Police heroin and Police cocaine. Both failed. In Exercise Four, the same pattern appeared with Police methamphetamine.

And yet, when presented with Customs samples — the same type the dog is routinely trained on — the response was immediate and clear.

So we have to ask:

Is the dog detecting the drug… or detecting its version of the drug?

Not all samples are the same. Even when they carry the same name, their odour profile can vary significantly depending on purity, cutting agents, source, storage, packaging and handling history.

If a dog is only ever trained on one type of sample, from one source, then what it learns is not “this is cocaine” or “this is methamphetamine”.

What it learns is:

“This is what it smells like in my training.”

And the moment that picture changes, recognition can fail.

Which is exactly what we saw.

The dog didn’t fail to search.

It didn’t lack motivation.

It didn’t miss the hides.

It simply did not recognise the Police samples as target odour.

But when presented with a familiar odour profile, it responded immediately.

That distinction is critical.

Because if training is built around a narrow odour picture, then what you create is not a detection dog.

You create a dog that recognises a profile, not a substance.

Storage, Handling and Sample Integrity

During these exercises, another factor stood out—and it’s something that cannot be ignored.

The way samples are stored and handled plays a critical role in what the dog ultimately learns.

Across different organisations, I regularly see significant variation in:

• Storage methods
• Container types
• Separation protocols
• Age and condition of samples

In this case, there were some concerns.

The drug samples were stored in jam jar–type containers, while firearms were stored in plastic lunchbox-style containers—all within the same case.

Now, on the surface, that may seem acceptable. The items are separated. They’re contained.

But in reality, this raises questions.

Because odour doesn’t respect containers in the way we often assume.

Cross-contamination is always a risk—especially when different odour sources are stored together, transported together, and repeatedly handled within the same confined space.

Added to that is the issue of sample quality and age.

Without clear control over:

• how long samples have been in circulation
• how they’ve been stored
• how often they’ve been handled

…the odour profile can change significantly over time.

So when we look back at the earlier exercises, another question emerges:

Was the dog failing to recognise the odour…

or was the odour itself inconsistent?

Firearms Sample Storage

What is particularly concerning is the practice of storing firearms in the same case as drug samples.

Even when items are contained separately, the shared storage environment raises legitimate questions around cross-contamination and odour integrity—especially over time.

This becomes even more significant when dogs are being trained to detect firearms.

These are not training props in isolation. Firearms are items that are actively smuggled across borders and used in serious criminal activity, including violence and loss of life.

If the odour picture associated with firearms is compromised through poor storage practices, then what the dog learns may not reflect the reality it will face operationally.

And that should concern all of us.

Why This Matters

If the training samples themselves are:

• contaminated
• degraded
• or inconsistent

then what we are presenting to the dog is no longer a clear, reliable picture.

And if the picture isn’t clear…

the learning won’t be either.

A Final Clarification

It’s important to be clear about something here.

This is not a criticism of the dog, the handler, or the service.

This dog can search. It is operational. It works as a Customs drug detection dog and carries out that role.

What we saw in these exercises does not take that away.

What this is, is an observation.

A moment to step back and look at what the dog showed us, without bias, without assumption, and without immediately trying to explain it away.

Because that’s where the value is.

These exercises weren’t designed to catch anyone out. They were designed to explore, to challenge, and to learn.

And that includes me.

I’m not writing this with all the answers.

I’m writing this because something didn’t add up, and instead of ignoring it, I chose to look at it properly.

To question it.

To understand it.

To learn from it.

Because that’s what we should all be doing.

Final Thought

We spend a lot of time focusing on indication, control, obedience and search patterns. But this scenario strips everything back to a single point.

Does the dog truly understand the odour it has been trained to detect?

Because if it doesn’t, everything else becomes irrelevant.

I keep coming back to that moment. The dog opens the box, places its nose directly inside, and walks away as if there is nothing there.

That isn’t just a missed find.

It’s a moment of truth.

And whether we like it or not, the dog has just shown us exactly where the problem lies.

So the question we should all be asking is this:

Are we training dogs to recognise odour…

or are we training them to recognise a version of it?