Zero data retention
Zero data retention (ZDR) is an account setting. While it is on, Spider runs your requests as usual and returns the results in the response, but keeps no page content, no request parameters and no cached copies. The request log keeps status, timing and credits, and holds each URL only as a keyed hash.
How to turn it on
Open account settings and switch on zero data retention. Only the account owner can change it, and in a team the team owner sets it. It covers new requests within about five minutes.
Enterprise customers can have it written into their agreement instead. Write to support@spider.cloud to arrange it.
Turning it off logs new requests in full again at the normal credit cost. Rows written while it was on stay hashed.
What Spider processes but does not keep
- Page content. Spider fetches the page, returns it in the response, and keeps no copy.
- Request parameters, such as headers, cookies, prompts and extraction schemas. Spider uses them to run the request and leaves them out of the log.
- URLs and hosts. The log holds a one-way hash in their place.
- Cached pages. Spider neither reads from nor writes to the page cache or any other cache for your requests, whatever the cache parameter says.
- User agent, error text and network addresses. The log row has none of them.
- Browser session recordings. Sessions you open still run, but Spider does not record them.
- Training data and samples. Spider uses nothing from your requests for training and takes no page samples from them.
The spider.cloud dashboard follows the setting too. It makes no session replay recordings of your visits, and the playground and the docs keep no run history in your browser.
What Spider keeps
| Record | What it holds | Kept for |
|---|---|---|
| Request log rows | Your account, the keyed hash, the status, timing and credits charged. | Up to 30 hours |
| Account and team data | Your email, sign-in method, API keys, team membership and settings. | While your account is open |
| Billing records | Credits bought and spent, with no URL or page content. | As long as tax and accounting law requires |
The abuse exception
If Spider refuses a ZDR request for abuse, it keeps a record of that request for up to 90 days. It never keeps the page content.
AI features and model providers
AI features, such as extraction with a prompt, send page content and your prompt to a model provider. For ZDR requests, Spider uses only provider endpoints that keep no data and do not train on it. If the model you ask for has no such endpoint, Spider uses a comparable model that has one.
Features that need storage
While ZDR is on, Spider turns off any feature that only works if it stores your results, for example to let you fetch them later. A request for one returns a 4xx error with the code zdr_unsupported and a reason, and the dashboard shows that reason in place of the result.
Streaming responses, webhooks to your own endpoint, and search keep working. A webhook payload goes only to the endpoint you configure, and Spider keeps no copy of it.
Pricing
On self-serve plans, every request costs 2.5x the usual credits while ZDR is on. Unlimited meters no credits, so the multiplier does not apply there. Enterprise agreements include ZDR at no extra charge.
The privacy policy covers how Spider handles data for every account.