Data freshness
CloudYali collects each provider's billing once a day, Cursor's every 6 hours, and reads recent days again each time. So the newest figures are a day or more behind, and can still change after they first appear.
Billing, per provider
| Provider | Collected (UTC) | First collection reads | Read again each day |
|---|---|---|---|
| AWS | About 04:30 | The current month | The whole current month, and the previous month until AWS invoices it |
| Google Cloud | About 07:30 | From the 1st of the month about 60 days back: two to three months | The last 15 days |
| Azure | About 13:40 | Up to 6 months of export files already in the container | The current and previous month |
| Anthropic | About 14:30 | 6 months | From the start of the last day collected |
| OpenAI | About 14:30 | 6 months | The last two to three days |
| Fastly | About 14:35 | Every invoice Fastly returns | The current month's invoice |
| Databricks | About 14:40 | 6 months, for your first workspace | Every row Databricks has added since the last collection, whatever its usage date |
| Cursor | Every 6 hours: 00:33, 06:33, 12:33 and 18:33 | 30 days of requests and 180 days of daily usage | The last two to three days |
The times are the earliest start. Several jobs run one after another, so a collection can start later on a busy day.
Nothing is collected the moment you connect a provider, except the first AWS resource scan. The first data arrives with that provider's next collection, plus any delay on the provider's side. AWS delivers the first Cost & Usage Report up to 24 hours after the stack creates it. Google Cloud's export starts on the day you turn it on.
Why recent days change
Providers keep adding and correcting charges after the usage, and each collection reads its window again: a day's AWS figure can change until the month is invoiced. Compare ranges that end at least two days ago when you match a provider's console.
A month AWS has already invoiced is not read again, so later AWS corrections to that month do not reach CloudYali.
Days and history
- Days. AWS, Microsoft Azure, Anthropic, OpenAI and Cursor use the day the provider reports. Google Cloud days are US Pacific days. Fastly cost is dated the 1st of each month.
- History. Cost Explorer holds about seven months of AWS, Google Cloud, Azure, Databricks and Fastly cost. See Cost types.
Other data
| Data | Refreshed (UTC) | Notes |
|---|---|---|
| Google Cloud resources | Once a day, about 11:30 | |
| Azure resources | Once a day, about 14:20 | |
| Databricks resources | Once a day, about 14:50 | |
| AWS resources | On a schedule; the first scan starts when you connect | |
| Commitments | Once a day, about 14:00 | From the billing data collected so far |
| Budgets | Once a day, about 14:55 | Spend from the start of the period to the latest data |
| Cost anomalies | Once a day, about 14:55 | For the day two days earlier |
| Cost Savings | Once a day, about 15:00 | Opportunities from every source |
| Daily email report | About 15:30 | For the day two days earlier |
| Cost Savings summaries to the owner | Mondays about 01:00, and the 1st of the month about 01:30 |
Where the console shows it
| Where | What it shows |
|---|---|
| Billing data through on Cost Views and Commitments | The latest billing day in the view |
| Last sync on the Connected tab of Integrations | The last collection. Google Cloud rows show —. For AWS it does not reflect cost collection, and for Azure it can be the resource sync |
| A budget's page | As of the last budget-monitor run |
| An anomaly's chart | Today is left out, because its data is still arriving |
CloudYali does not show a failed collection in the console. If data stops arriving, see Data is not showing.