IoT in Retail: Sensors, Smart Fridges and Edge Telemetry in Real Stores

IoT in retail is the network of connected devices — smart fridges, AI cameras, electronic price tags and door controllers — that lets a physical store sense, report and operate without staff on site. Each device is a telemetry endpoint feeding a central system, and in an unmanned store that telemetry is not a nice-to-have: it is the only way anyone knows the store is working.

We see this directly through StoreTech, the unmanned retail platform Things at Web operates. StoreTech's technology powers 24-Sju, a chain of unmanned grocery stores across Sweden focused on rural communities, where customers enter with BankID and check out by self-scan. Building for that environment forces a particular discipline, and this article walks through the IoT engineering it demands.

The store as a fleet of IoT endpoints

A staffed shop hides a lot of state in the heads of the people working there. Remove the staff and every one of those signals has to become a measurable, connected device. On the StoreTech platform the core endpoints are:

• Smart fridges — monitored cold storage reporting temperature and, with integrated sensing, what is taken out.
• AI cameras — computer-vision sensors for loss prevention, occupancy and incident detection.
• Electronic price tags (ESL) — wirelessly updated shelf labels so pricing changes propagate centrally, never by hand.
• Door control systems — the access layer that gates entry, in 24-Sju tied to BankID identity.
• Shopping kiosks and self-scan checkout — the customer's transaction point.
• Vending machines — additional automated dispensing endpoints.

All of these are listed as StoreTech modules on store-tech.se, tied together with ERP and mobile apps. The engineering insight is that each one is simultaneously a control surface (you command it) and a sensor (it reports back). Treating them as a single managed fleet, rather than separate gadgets, is what makes a store operable from a screen hundreds of kilometres away.

Connectivity and edge in rural Sweden

24-Sju's rural focus is the hard part. A store in a small community may sit on a single consumer-grade broadband line or a cellular link, with no IT staff and no quick on-site fix. That shapes the architecture.

Why the edge matters

You cannot assume the cloud is always reachable. Latency-sensitive and safety-relevant functions — unlocking a door for an authenticated customer, letting checkout proceed, keeping a freezer compressor controlled — must keep working through a connectivity drop. So the controlling logic lives at the edge, on local hardware inside the store, with the cloud used for coordination, analytics and oversight rather than as a single point of failure.

A practical split looks like this:

Layer	Runs where	Responsible for
Device	The fridge, camera, lock, tag	Sensing and actuation
Edge	Local controller in the store	Real-time decisions, buffering, local control loops
Cloud	Central platform	Fleet telemetry, dashboards, OTA, ERP, analytics

When the link drops, the edge keeps the store running and buffers telemetry; when it recovers, buffered data syncs upstream. Designing for the disconnected case first, then layering cloud on top, is the order that survives contact with a rural site.

Telemetry dashboards and OTA management

Operating 130+ stores with no staff means a small central team needs to see everything and act remotely. Two systems make that possible.

Telemetry dashboards aggregate the fleet's signals into one operational view: fridge temperatures trending toward a threshold, a camera offline, a door controller reporting faults, a kiosk that stopped taking transactions. The goal is to move from reactive ("a store called us") to predictive ("this compressor's temperature curve says service it before it fails"). Good telemetry is structured, timestamped and continuous, not a log you read after something breaks.

OTA (over-the-air) management is non-negotiable at fleet scale. You cannot send a technician to every rural store to patch firmware. OTA lets you push updates and security patches to devices remotely — but it demands rigour: staged rollouts, the ability to roll back a bad update, and a guarantee that a device interrupted mid-update recovers rather than bricking. A failed OTA on a door controller in an unmanned store is not a bug ticket; it is a locked-out customer with no one to help them.

What unmanned operation demands of reliability

This is the theme under everything above. When there is no one on site, device reliability stops being a quality metric and becomes the business model.

A few principles we apply:

• Fail safe, not fail open. Define the safe state for every device when it loses connectivity or power, and default to it.
• Degrade gracefully. A camera going offline should not stop the door from opening for a paying customer; isolate failures so one device does not cascade.
• Make every device observable. If you cannot see a device's health remotely, you cannot operate it remotely — and you will be dispatching people, which defeats the model.
• Design for the worst site. Architect around the store with the flakiest connection and the longest drive time, not the flagship.

Get those right and an unmanned store is not a constant fire drill; it is a managed system. If you are building connected hardware or the platform behind it, our IoT solutions and AI & machine learning teams work on exactly these problems — see the StoreTech platform or get in touch.

FAQ

What does IoT mean in a retail context?

In retail, IoT means connecting physical store devices — fridges, cameras, electronic price tags, door controllers, checkout kiosks — to a central system so they can be monitored and controlled remotely. Each device reports telemetry and accepts commands, turning a physical store into a measurable, manageable system rather than something that depends on on-site staff.

How do unmanned stores work without staff?

Unmanned stores like 24-Sju replace human tasks with connected devices. Door control tied to BankID handles secure entry, AI cameras handle loss prevention and monitoring, self-scan kiosks handle checkout, and electronic price tags handle pricing. A central platform watches all of it via telemetry, so a small remote team can operate many stores at once.

Why is edge computing important for smart stores?

Edge computing runs critical logic on local hardware inside the store rather than depending on the cloud. This matters because rural stores can lose connectivity, and functions like door access, checkout and refrigeration must keep working through an outage. The edge handles real-time decisions and buffers data; the cloud handles oversight, analytics and updates.

What is OTA management and why does it matter?

OTA (over-the-air) management is the ability to push firmware and security updates to devices remotely, without sending a technician on site. For a fleet of unmanned stores it is essential, but it requires staged rollouts, rollback on failure, and recovery if an update is interrupted — because a failed update on a door or fridge in an unstaffed store has direct customer impact.

What makes device reliability so critical in unmanned retail?

With no staff on site, a device failure cannot be patched over by a person. Reliability becomes the business model: every device needs a defined safe state, failures must be isolated so one fault does not cascade, and every device must be observable remotely. Architecting around the worst-connected site keeps an unmanned store a managed system rather than a recurring emergency.