A logistics platform receives location updates from a large fleet every few seconds. Where should volume be reduced, and what is the ordering of leverage?

The ordering of leverage

1. Filter at the device. Transmit on meaningful change — moved beyond a threshold, changed heading, changed state — plus a low-frequency heartbeat. The largest available reduction and it happens at the cheapest possible point, before anything is transmitted, stored or processed — and it saves the device's metered bandwidth and battery too.

2. Conflate on the live path. Where updates supersede, keeping only the latest value per device bounds the work by fleet size rather than by message rate, converting an unbounded problem into a bounded one.

3. Separate paths by consequence. Live tracking wants the latest position and can drop the rest; route reconstruction wants completeness and tolerates minutes; analytics wants aggregates. One pipeline for all three pays the strictest requirement of each — completeness and low latency simultaneously — which is the expensive combination.

4. Aggregate before the expensive stages, since most consumers need "where is it now" and "what path did it take" rather than every point at full fidelity.

5. Tier retention: full fidelity briefly, downsampled longer, compressed archival.

What must not be conflated

Events where each one matters: a fault code, a delivery confirmation, a trip completion, an update acknowledgement. Conflating those is data loss, and the distinction must be explicit in the design rather than a property of the transport — which is exactly why they belong on a different path.

The protocol dimension

A persistent connection with a lightweight keepalive beats reconnecting per message, since a handshake on a mobile network costs a significant fraction of a second and disproportionate power. And quality-of-service applied per message type — fire-and-forget for a position, acknowledged for a fault — since one level for everything either wastes bandwidth or loses important events.

The unit metric

Cost per device per day, trended. Total cost rises with the fleet and tells you nothing; unit cost falling as the fleet grows is the only evidence the efficiency work is compounding.