IoT fleet TCO is bigger than the device BOM.

1. Separate acquisition from operation

Start with hardware, provisioning and installation as acquisition costs. Then model connectivity, cloud/platform and operational support as recurring costs. Keeping them separate makes it easier to see what scales with new deployments versus what follows every active device.

2. Put field work into the model

A small annual failure rate can create a large fleet-level cost once technician time, travel and replacement activity are included. Model an annual failure percentage and an average cost per field intervention instead of assuming failures are exceptional.

3. Model fleet growth explicitly

A 25% annual growth assumption affects more than hardware purchases. It increases monthly connectivity, cloud load, support exposure and the absolute number of field failures. The operating base compounds.

4. Stress-test, do not just forecast

Create at least three cases: a base case, a downside case with more expensive field/network assumptions, and an upside case based on validated production pricing. A single forecast can hide which assumption actually drives the result.

5. Add telemetry and battery once the cost model is stable

Message frequency, payload size and radio behaviour influence data cost and energy use. A commercial model becomes stronger when those engineering assumptions can be connected to operating economics — while still being validated against real devices and supplier tariffs.