Trello board recovery: JSON fidelity with scripted restore versus CSV simplicity with manual reconstruction for sub-four-hour RTO

Answer first

For a 2,000-card board with checklists, comments, custom fields and attachments, a scripted JSON restore via Trello API v2 is the only path that can meet a 4-hour RTO with fidelity. CSV import is fast for basic card fields but cannot meet the fidelity requirement and the manual reconstruction cost is not feasible within 4 hours.

Confirmed: Trello API v2 JSON export captures board, lists, cards, checklists, comments, custom field values and attachment metadata. CSV export via UI omits checklists, comments, custom fields and attachment records. Custom field definitions are workspace-scoped. Trello Enterprise offers 30-day deleted-board recovery, not point-in-time restore for individual boards. Rate limit is 300 requests per 10 seconds per token.

Likely: A 2,000-card board needs ~14,000+ API calls for full fidelity restore. Pure API time at 30/s sustained is ~8 minutes, but dependency ordering, retries, pagination and attachment handling push realistic scripted restore to 2-4 hours for a well-tuned script. That leaves minimal margin for a 4-hour RTO and assumes no failures. CSV manual reconstruction of checklists, comments, custom fields and attachment metadata for this size is weeks of effort, not hours.

1. Can scripted JSON restore finish within 4 hours?

Likely yes, with conditions. Confirmed facts do not guarantee timing; actual duration depends on script efficiency, error handling and attachment volume.

Likely explanation: Dependency chain is board → lists → cards → checklists → checklist items → comments → custom field items → attachments. Board-level operations serialize on board ID, limiting parallelism gains. Attachment files are not in JSON, only metadata/URLs, so file re-upload is extra work if files are lost.

Steps needed for this case:

• Run a 200-card test mirror with same checklist, custom field and comment density. Export JSON, restore via script with single token, measure end-to-end time and call count. Scale linearly to estimate 2,000-card time.
• Implement rate-limit handling with X-RateLimit-Remaining monitoring and exponential backoff. Avoid burst-only logic.
• Pre-create lists and custom field definitions in target workspace before card creation to reduce dependency waits.

2. CSV speed vs manual reconstruction cost

Confirmed: CSV import is UI-simple for name, description, due date, labels. Checklists, comments, custom fields and attachment metadata are not present.

Likely: Reconstructing ~5,000 checklist items, ~2,000 comments, ~4,000 custom field values and ~1,000 attachment records manually is not achievable in a 4-hour RTO. Third-party backup tools such as Rewind, Backups for Trello and SpinBackup offer automated point-in-time restore with higher fidelity than CSV, but restore speed remains bound by Trello API rate limits and they add cost and vendor dependency.

3. Restore to a different workspace and custom field remapping

Confirmed: Custom field definitions are workspace-scoped and API does not automate remapping.

Likely: Remapping does not negate JSON fidelity advantage. It adds ~30-60 minutes of script logic to recreate field definitions in the target workspace and map source field IDs to target IDs before writing custom field items. CSV plus manual field recreation forces the same remapping work per card manually, which is far less predictable.

Steps needed:

• Export field definitions from source workspace and recreate equivalent definitions in target workspace with matching type and options.
• Build an ID mapping table source_field_id → target_field_id and apply it during restore.
• Validate field values populate on a sample of cards before full restore.

Missing diagnostic detail that changes recommendation: Is the restore target the same workspace as the source? Same-workspace restores avoid custom field remapping and reduce risk. If cross-workspace is required, confirm that equivalent custom field definitions can be pre-created in the target workspace before the restore window starts.