Skip to main content
The amount of historical data a connector can show is set by two independent caps:
  1. The source platform’s retention — how far back the platform’s API will return data.
  2. The initial backfill window — how much history is pulled on the first connection.
When a widget requests a date range that exceeds either cap, the missing portion comes back empty without an error.

Adriel keeps history beyond the source platform’s window

For cached connectors, every daily snapshot Adriel pulls is retained. This matters because most ad platform APIs trim how far back data can be re-fetched. For example:
  • Google Search Console only serves about 16 months from its API.
  • Naver Search Ads keyword-level data is limited to 180 days per request.
  • Amazon Vendor Central caps API access at 180 days.
  • Amazon DSP caps at 60 days for first-time fetches.
Once Adriel has pulled and cached a day, it stays available in the dashboard — even after the source platform stops returning that day from its own API. The practical effect: the longer a connector has been continuously connected, the deeper its historical view becomes inside Adriel, often well beyond what the source platform would still return today. This is one of the under-recognized benefits of long-term Adriel use. A workspace connected to Google Search Console two years ago can still query month 23 in a dashboard, even though re-fetching that month from Google’s API would now be impossible.

How much history is available?

The short answer: up to 2 years for most ad platforms, but the real cap depends on the connector and how long it’s been connected. Some examples: For the initial backfill and rolling refresh window of any connector, see its page under Data Sources.

Initial backfill vs ongoing refresh

When a data source is first connected, the connector runs an initial backfill — typically 60 to 180 days. After that completes, the connector switches to its rolling refresh window (3 to 30 days), which only updates the trailing window plus today. Days outside that window are not re-pulled automatically. The implication: if a connector’s initial backfill is 90 days, that’s the starting point. Once the data source has been connected for longer than the initial backfill window, history grows day by day, up to whatever cap the source platform allows at fetch time — and persists in Adriel’s cache after that.

What happens when a date range exceeds the cap?

Querying a date range that pre-dates the available history returns empty rows for the unavailable portion. No error is shown; the widget simply renders fewer data points than expected. To check whether the date range is the cause, narrow it to a known-good window and confirm the data appears.

Troubleshooting

Adriel can only retain data it has previously fetched. For days before the data source was first connected, history is only available if the source platform’s API still returns those days — which becomes less likely the further back the range goes.
The initial backfill only fetches a limited window (typically 60 to 180 days). Days older than that window only exist if the data source has been connected continuously since then. Run a manual refresh covering the missing range from the connection’s Refresh tab if the connector supports it.
Expected for new connections. Google’s API retains roughly 16 months of search performance data. A workspace connected longer than that retains the cached older data inside Adriel even after Google stops returning it.
DV360 enforces per-breakdown caps on its API: 93 days for the daily breakdown, 1 year for weekly, 2 years for monthly. Switch the widget to a weekly or monthly breakdown for older ranges. Days already cached at daily granularity remain available inside Adriel.
Deleting a data source is a hard delete; there is no self-serve recovery. Recovery is only possible as an exception where Adriel operations restores from a database backup, which is best-effort and comes with operational cost. Contact support if recovery is required.
Related: Refresh strategies · Time zones & data freshness