Data freshness
How current each provider's data is in OpsLyft, and how incomplete days are marked.
Providers publish billing data on a delay. OpsLyft shows you how current each source is so you don't compare a complete day with an incomplete one.
Where freshness is shown
| Place | Indicator |
|---|---|
| Clouds picker in Cost Explorer | live or Nh behind per provider |
| Cost Explorer toolbar | LIVE dot |
| Atlas headers | Inventory · polled 30 h ago, Billing through Sep 25 |
| Change history | Change log · polled · 30 d kept |
| Anomalies | Detector ran Sep 25, 19:19 |
| Iris answers | A freshness note on the date range |
| Tickets | Jira synced 2 min ago |
Typical delays
| Provider | Typical delay |
|---|---|
| AWS | A few hours for new data; daily totals settle over about 48 hours |
| Azure | 8–24 hours, longer around month-end |
| Google Cloud | A few hours to a day |
| Snowflake | Up to a few hours |
| Databricks | Same day |
| OpenAI, Anthropic | Hours |
| Kubernetes | Near real time from the agent |
Month-end credits, refunds, and invoice adjustments can land several days after the month closes.
Reading recent days
- Treat the last one to two days as provisional.
- When comparing providers, check the clouds picker; a provider 71h behind will look low for its last three days.
- Iris tells you when a period falls inside the usual billing lag.