Why most unmanned retail concepts don’t scale.
By Jan Marck Vrijlandt – Co-founder & Board member MAIKOZ
Over the past years, I’ve seen many unmanned retail concepts come to life. At the start, almost all of them look successful.
The hardware works.
Customers engage.
Transactions come in.
Dashboards look promising.
At one or two locations, everything feels under control. Then expansion starts.
And that’s usually where things begin to break.
Not because demand disappears.
Not because the technology fails.
But because the system was never designed to scale.
Scaling is not adding locations
One of the most persistent misconceptions I see is this: people think scaling means adding more units. It doesn’t. Scaling means your structure holds under pressure. If adding locations increases complexity, cost, or variability, you don’t have scale. You have expansion.
The illusion of early success
A pilot location can prove demand. But it doesn’t prove scalability. At small scale, people compensate manually:
Someone checks inventory daily
Someone adjusts pricing case by case
Someone steps in when something breaks
At 1–2 locations, this works. It even feels controlled. But manual effort is masking structural weakness. As soon as you grow, those weaknesses surface. If your concept depends on constant supervision, it’s not scalable.
What scalability actually means
For me, scalability is quite simple. It means:
Performance remains predictable
Service costs stay under control
Margins don’t erode as you grow
Complexity does not increase per location
Scaling should create efficiency, not friction.
Where I see most concepts break
Across markets, I keep seeing the same patterns.
1. Complexity increases with every location
Most concepts start simple. Then exceptions creep in. Different assortments. Custom pricing. Extra SKUs. Before long, the system becomes difficult to manage. What felt tailored in the beginning becomes unmanageable at scale. Simple systems scale. Complex systems don’t.
2. Data is collected, but not used
Almost every setup today collects data. But data alone doesn’t create scalability. What matters is whether it drives decisions. In many cases, it doesn’t. If insights require manual interpretation, scaling slows down. Reactive systems don’t scale. Predictive systems do.
3. Service is not engineered
Unattended retail is still a physical operation. Products need to be refilled. Machines need maintenance. Issues need to be resolved. If your service model is based on reacting instead of structuring, costs become unpredictable very quickly. At scale, that becomes unsustainable.
4. Compliance risk grows quietly
At a small scale, compliance is often handled manually. At scale, that becomes exposure. If compliance depends on people, risk increases with every location. It should be built into the system, not added on top.
5. Unit economics are assumed, not understood
Another pattern I see often: “If it works here, it will work everywhere.” That’s rarely true. Every location behaves differently. Scaling without understanding that variability is not strategy.
It’s speculation.
The real issue
Most concepts are designed to operate. Not to scale. That’s where it goes wrong.
How I approach this at MAIKOZ
At MAIKOZ, we don’t treat scalability as something that comes later. It’s something we look at from the start. What matters to us, and what we actively help partners achieve – is a system that absorbs complexity.
That means:
reducing manual decision-making
building in structure before growth
making performance visible in real time
ensuring compliance doesn’t rely on individuals
validating the economics before expanding
The operator shouldn’t carry the system. The system should support the operator.
Why this matters now
Interest in unmanned retail is increasing rapidly. Technology is improving. Consumers are adapting. Labour costs are rising. There is a real opportunity.
But I also see the same mistake repeating: speed before structure. The market doesn’t need more pilots. It needs scalable systems.
Final thought
If your concept:
requires constant supervision
becomes more complex with each location
depends on manual decisions
lacks real-time visibility
has unclear downside economics
Then it’s not scalable.
Scaling is not confidence.
Scaling is engineering.
