Where the cost comes from
Telemetry has costs at both ends of the link, and only one end sends an invoice.
On the device, every event takes CPU time to record, storage to hold until it can be sent, and radio time to upload. For a phone that means battery and, on cellular, the user’s data allowance. For a sensor on a metered link it means airtime someone pays for. These costs never appear on an observability bill, but they grow with every event you choose to send. Sending in batches rather than one event at a time cuts the radio overhead; see why batching saves battery and bandwidth.
In the pipeline, someone pays to receive the data, process it, store it, keep it for some period, and run queries over it. Whether you self-host or buy a service, those are the same five activities; a service turns them into a price.
How vendors price telemetry
Providers use a handful of units, and some combine several. Their own pricing pages show the pattern:
- Per gigabyte ingested. New Relic says its pricing rests on users and data ingest, with a free monthly ingest allowance and a rate per GB beyond it. Google Cloud Logging charges per GiB when logs are streamed into log bucket storage, and that charge includes up to 30 days of storage.
- Per gigabyte at each stage. Grafana Cloud prices logs and traces separately per GB processed, per GB written and per GB retained.
- Per event stored. Datadog meters its Flex Logs tier per million events stored, alongside per-GB charges for ingesting and forwarding.
- Per metric series. Grafana Cloud prices metrics per thousand active series, so the count of distinct label combinations, not only the volume of data points, drives the cost.
- Per user or seat. New Relic charges for full platform users as well as data.
- Retention. Google Cloud Logging charges per GiB per month for logs kept longer than 30 days; other vendors tie retention to the plan or sell longer retention as an option.
- No direct charge. Firebase lists Crashlytics and Performance Monitoring as no-cost products.
The unit changes which events are expensive. Under per-GB pricing, a verbose event with large attributes costs more than a compact one. Under per-event pricing, a tiny event costs the same as a large one, so splitting one record into several raises the bill. Under per-series pricing, an attribute that takes a new value per device or per user can multiply the series count. Read what one billable unit is before comparing rates.
What drives the bill
Whatever the unit, the same factors set the total:
- Devices that report.
- Events per device, set by what you instrument and how often.
- Size of each event, set by the attributes you attach.
- Retention, how long the data is kept in a queryable state.
- Duplicates, if retried uploads are counted twice. A unique identifier on each telemetry event lets a backend discard copies.
The cheapest event is the one you never send. Sampling telemetry keeps a representative fraction of routine events; aggregating on the device turns a stream of readings into one counter or histogram; and short retention for detailed data, with summaries kept longer, cuts storage.
One published example
Offline Protocol publishes its telemetry prices. The Free, Pro and Scale plans cost $0, $99 and $499 a month and include 250,000, 2M and 10M telemetry events a month respectively. Events beyond the quota cost $0.05 per 1,000, and Enterprise volumes are contracted. The meter counts events accepted by the ingest, and the plans show 7, 30 and 90 days of analytics history.
With a card on file, usage beyond the quota is billed at that rate, and you can set a billing limit on each meter so charges for it never exceed the limit. A Free organization without a card stops at its quota instead: the first meter to reach it pauses all of the organization’s metered hosted services until a card is added, a paid plan is chosen, or the next billing period starts.
Telemetry there is opt-in. The SDK sends nothing until the app calls enableTelemetry with a key and App ID. That key ships inside the app, and the security page lists the risk plainly: a key embedded in an app can be extracted, so scope it, monitor it and revoke it if it is abused. Anyone holding it can send telemetry that counts toward your usage, which is a second reason to set a billing limit on the telemetry meter.