Supply chain visibility: why the data layer comes before the dashboard
How IoT sensors and real-time data pipelines are changing what companies can know about their supply chains - and where most projects stall.
After two years of global supply chain disruptions, visibility became the word every operations director was saying. Track shipments in real time. Know where inventory is. React before the shortage hits the floor. The business case is obvious. The technology projects launched in response to it have had a much harder time delivering.
I have worked with a handful of these projects over the past year and I want to share a pattern I keep seeing: companies focus on the output - the dashboard, the alert, the insight - and underinvest in the layer that makes those outputs possible.
What supply chain visibility actually requires
At the infrastructure level, real-time supply chain visibility requires three things to be working together: sensors or data sources at the edge, a reliable pipeline from those sources to somewhere central, and a data model that makes the information queryable across locations, vendors, and time.
Companies typically have partial versions of each. Some parts of the chain are instrumented, others are not. The pipeline works for some data sources and falls back to manual entry for others. The data model was designed for one system and does not cleanly represent what was acquired or built later.
The dashboard sits on top of this inconsistent foundation and shows numbers that operators do not fully trust - because they know which parts of the chain are reporting in real time and which are being estimated.
Where IoT devices actually fit
IoT sensors on pallets, containers, or machinery generate one thing: a stream of timestamped readings from a specific physical location. Temperature, GPS coordinates, door open/close, weight, vibration - depending on the device.
That data is only useful when it is joined to context: which product is in that container, which order it belongs to, which customer it is going to, what the expected transit time is. The sensor data alone is noise. Combined with the operational data, it becomes signal.
This is why the data layer matters more than the device. A company can instrument every truck in its fleet, but if the sensor readings cannot be reliably joined to the order management system and the inventory records, you end up with a map that shows trucks but cannot tell you whether the right goods are on them.
The integration problem that kills these projects
Most supply chain visibility projects that I have seen struggle at the integration layer. The data from sensors goes into one system, the ERP holds the order data, the WMS holds the inventory data, and there is no single place where these are joined and maintained.
The result is that generating a useful report requires a custom query written by someone who knows the quirks of all three systems. That person exists, usually in the IT department, and becomes the bottleneck. The dashboard that was supposed to give operations directors independence ends up creating a dependency on one engineer who understands the joins.
The fix is not glamorous: it is a properly modeled central data layer - sometimes called an operational data store or a supply chain data hub - where the records from ERP, WMS, and IoT sensors are unified into a consistent model. This is data engineering work, not sensor work and not dashboard work.
What to look for in a project proposal
When a vendor or an internal team proposes a supply chain visibility project, the useful questions are:
- How many separate data sources feed this, and who manages the integration for each?
- Is there a data model document that shows how an order traces from source to delivery through all the systems involved?
- What happens when a sensor goes offline or misses readings - how does the system represent that gap to the user?
- How will the dashboard be maintained when the underlying ERP or WMS changes?
If the answers are vague on the integration and data model questions, the proposal is still in the demo stage. The real work has not been scoped yet.
The practical starting point
For companies that are early in this, I recommend starting with a data audit before any sensor purchase. List every system that holds data about your supply chain - orders, inventory, transport, suppliers - and map what exists, who owns it, and what the data quality is like. That audit will tell you more about what your visibility project actually needs than any vendor demo.