## TL;DR

The agent moved billing data to Deep Archive to save storage costs, but the auditor needs standard (fast) retrieval, and Deep Archive's hours-long retrieval breaks the audit timeline. Restore the audit-relevant billing data to a standard-retrieval tier, and set the archival policy to keep audit-window data in standard retrieval. Retrieval speed is part of the data's value.

## Error

```text
agent archived the billing data to deep archive  -  the auditor needed standard retrieval
```

## Steps

1. Confirm the audit requirement: the retrieval SLA the auditor needs. Expected: the required retrieval speed is documented.
2. Restore the needed billing data from Deep Archive to standard storage (initiate the restore early; it takes hours). Expected: the restore is in progress.
3. Provide the auditor interim access to whatever is already in standard tiers. Expected: the audit can proceed on available data.
4. Change the archival policy: billing data within the audit window stays in standard retrieval; only older data goes deep. Expected: the policy respects the audit SLA.
5. Document the retrieval-tier decision so future cost runs do not re-archive audit data. Expected: the constraint is recorded.

## When to use

- Billing data is archived too deep for audit retrieval needs.
- The auditor faces hours-long restores.
- Setting archival tiers for compliance data.

## When not to use

- The data is beyond any audit window (deep archive is fine).
- No audit is planned or required.
- The issue is storage cost, not retrieval speed.

## Tool compatibility

AWS S3 storage classes, Glacier (current 2026 releases).

### Variant phrasings

### auditor waiting on deep archive restore

Restore to standard and fix the tiering policy.

### billing data archived too aggressively

Keep the audit window in standard retrieval.

### retrieval sla vs storage savings

Speed is part of the data's value.

## Why it happens

Archive tiers trade retrieval speed for storage cost, and the tradeoff looks pure-win until someone needs the data fast. Audit timelines assume standard retrieval; Deep Archive's multi-hour restores break them. The cost-optimal tier for data nobody reads is the wrong tier for data an auditor reads on a deadline. Tiering policy must encode who reads the data and how fast, not just how old it is.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_S9ngfazF51MKBg22JKjrww