“Cloud accounting” usually means someone else hosts the database. “Local-first” turns that around: the primary ledger file is yours, operated by software that runs close to you. In Privbooks, that means SQLite — a battle-tested embedded engine — living in the browser with modern storage APIs where available.
What you can export — and why it matters
A normalized .sqlite3 file is not a pretty PDF; it is a portable artifact you can copy, version, and attach to your disaster-recovery thinking. Compare that to “export a CSV of reports” — helpful, but not equivalent to owning the relational truth of your books. Reports are derived; the database is the spine that produces them consistently.
- CSV exports are great for CPAs, pivot tables, and one-off analysis.
- Full database export is the continuity story when you change devices or verify integrity.
- Preparer ZIPs bundle operational schedules; still not a substitute for the database when you need to re-open history.
Browser context without hand-waving
Running SQLite in a tab has constraints: storage quotas, security sandboxes, and cross-origin isolation affect durability and performance. Serious local-first apps document those expectations instead of promising desktop parity everywhere. Practical takeaways:
- Durability improves when the app can use private origin storage features exposed by modern browsers under the right deployment headers.
- You still manage backups—browser storage is not a substitute for an operator’s retention policy.
- Migrations and cutovers are when export discipline matters most; see QuickBooks cutover playbook.
A backup cadence that matches risk
A useful default is weekly for active books, daily during month-end crunch, and immediately before browser/OS upgrades or device swaps. Treat exports like insurance: cheap when unnecessary, priceless when necessary. Pair exports with a simple naming convention (company-YYYY-MM-DD.sqlite3) and an off-device copy (encrypted disk, object storage, or your firm’s document policy).
Evidence, not theater
Lenders and buyers ask for consistency: registers reconcile to statements, subledgers roll into the trial balance, and supporting detail ties to invoices. A portable ledger file supports that story when paired with your process — not when substituted for it. If you are evaluating lock-in risk more broadly, read data ownership vs hosted GLs.
Try it with a deliberate drill
The best time to learn recovery is before you need it: export, restore into a test profile, and verify reports match. If that workflow feels sane, you have a continuity story. Start with Privbooks if you want double-entry machinery where the default posture is a file you can hold.