Why the measure is per call
The cost is per call and the work is per call. Two numbers on the same axis are comparable.
The honest outcome
A design that is refused. The refusal is written with the two numbers, and the write-up is the deliverable.
A module costs startup, memory, and crossings. If the work inside is small, those costs are larger than the work, and the design is a loss rather than a win.
About 1 min of reading
Compare work to overhead
Write the two numbers
Before you run it, say out loud what it will print. Then open the answer.
Amortise the overhead
Before you run it, say out loud what it will print. Then open the answer.
Write the decision
Before you run it, say out loud what it will print. Then open the answer.
The cost is per call and the work is per call. Two numbers on the same axis are comparable.
A design that is refused. The refusal is written with the two numbers, and the write-up is the deliverable.
Instantiating a module is work. A loop that instantiates per item pays that work per item.
The module reserves a flat array. The reservation is the footprint whether or not it is touched.
The question is not whether WASM runs. The question is whether it beats what runs natively.
Each one is defined here on this site. Open the entry for the full definition and the other words it points at.