Customers evaluating FIELDKIT usually start by comparing unit price against alternatives. That comparison is nearly meaningless, and I would rather explain why than win a deal on it.
Over a five-year deployment, the node hardware is typically 20–25% of total cost. The rest is site selection and permission, installation labour, connectivity, maintenance visits, calibration, and the cost of the data being wrong for a period before anyone notices.
That last item is the one nobody models and the one that dominates. A node that fails silently in month seven of a twenty-four month deployment does not cost you a node. It costs you seventeen months of a gap in a time series you were using to make decisions.
This is why the 400 spends power on heartbeats and self-diagnostics that could have been spent on battery life, and why the app nags about overdue calibration. Availability of correct data beats availability of data.
The deployment planning guide on the downloads page has the full cost model, including the parts where the numbers are unflattering to us.