Service levels
What we commit to, in writing
Every plan carries written service levels: when data runs, how fast a break gets fixed, and what happens when a delivery is late. These go in your order form, not just on this page.
Refresh schedule
Each plan runs on the agreed cadence — weekly, daily or hourly — at a fixed time in UTC. Every row we deliver carries a timestamp, so you always know exactly when it was collected.
Fix targets
When a source breaks — a layout change, a new bot check, a blocked page — our checks catch it automatically. Repair starts the same business day. The target is a fix the same day on Scale, and by the next business day on Monitor.
Delivery
Data lands in your agreed target within the delivery window for your plan, with a success notification. If a run is going to be late, we tell you before the window closes, not after.
Schema stability
We add fields; we don't repurpose or silently remove them. Any field removal is announced at least 30 days ahead. Every row carries a schema_version column so your pipeline can handle changes safely.
Data quality checks
Before any delivery goes out, we check row-count variance against the last run, the empty-field rate, and duplicate rows. A failed check holds the delivery and notifies you rather than shipping data we haven't verified.
By plan
| Plan | Refresh | Fix target | Delivery window | Support |
|---|---|---|---|---|
| One-off dataset | Once | n/a | Agreed date | |
| Monitor | Weekly or daily | Next business day | Within 24h of scheduled run | Email, 1 business day |
| Scale | Daily or hourly | Same business day | Within 4h of scheduled run | Slack, same day |
| Your data team | As agreed | Same business day | As agreed | Dedicated engineer in your Slack |
When something breaks
How we classify
We sort incidents into three kinds: a blocked source (access denied or a bot challenge), a layout change (the page structure moved and extraction broke), and a delivery failure (the run completed but didn't reach your target).
What you'll hear from us
When something breaks, you get a notification, an ETA for the fix, and — once it's resolved — a short note in our repairs log. We tell you what happened in plain terms, not a status code.
Credits
A missed delivery window on Monitor or Scale credits that run's share of the month. The terms are written into your order form, not left to a case-by-case decision.
Service level questions
What counts as a missed delivery?
A run that doesn't land in your target within your plan's delivery window.
Do fix targets apply to every source?
They apply to sources we're actively monitoring under Monitor, Scale or Your data team. A one-off dataset has no ongoing fix target, since there's no recurring run to repair.
Can we get a different fix target or delivery window?
Yes. On Scale and Your data team we can set different terms in the order form.
Need these terms in a contract?
We attach service levels to every order form.