A data platform is judged by the decisions it carries
Warehouses get measured on uptime and cost. The number that matters is how many decisions moved because the data arrived in time.
Platform reviews tend to open with infrastructure metrics. Uptime, query latency, monthly spend. All are worth knowing and none of them say whether the platform is doing its job, which is to put the right number in front of the person who has to act on it, early enough for the action to matter.
Three questions that expose the gap
- Which recurring decisions depend on this data, and who makes them?
- How stale is the number at the moment of the decision, not at the moment of ingest?
- When the pipeline breaks, who finds out first — the platform team or the decision maker?
The third question is usually the revealing one. If the answer is the decision maker, the platform is not observable enough to be trusted, and people will quietly build a spreadsheet beside it.
Real-time is a requirement, not a posture
Real-time architecture is expensive to run and expensive to staff. It is justified where the value of a decision decays in seconds or minutes — trade surveillance, fraud, pricing, operational alerting. Everywhere else, hourly batch with honest freshness indicators serves people better than a streaming system nobody can debug at two in the morning.
Built to carry decisions, not just store records.
A platform review that starts from the decision list, rather than the schema, usually finds that a third of the pipelines feed nothing at all.