What data we collect
This page is a plain-English inventory of what Product Research stores about you. The deeper retention/policy details are in Retention & policies.
In this guide:
- Per-search snapshots
- Watchlist
- Chat sessions
- AI briefs
- Exports
- Credit ledger
Per-search snapshots
When you run a search, Hilal stores:
- The search parameters (keyword, marketplace, category, filters, weights).
- The results — frozen at the moment of search. ASIN, title, brand, price, BSR, reviews, rating, all six sub-scores, the composite SellerScore, data-quality flags, and warnings.
- The user ID and organization ID from your Hilal Auth token.
- Credits charged for audit purposes.
- Timestamps (created, updated, deleted).
Tables: research_searches, research_products_snapshot in Postgres.
Why: History viewing, re-fetch, re-run, and export all need this data.
Watchlist
When you add a product to the watchlist:
- The product details (ASIN, marketplace, cached metrics).
- Your free-text notes.
- Optional source-search UUID linking the watchlist entry back to the search that surfaced it.
- Created and last-refreshed timestamps.
Table: research_watchlist.
Why: Watchlist viewing, refresh, notes editing.
Chat sessions
For every chat session:
- Session metadata (UUID, title, your user ID, organization ID, created/updated timestamps) in Postgres.
- Message bodies — your messages and Claude’s responses, including any tool-use payloads — in MongoDB.
- Reasoning logs for transparency / debugging — also in MongoDB.
Tables: research_chat_sessions (Postgres), messages (MongoDB).
Why: Conversation persistence, session resume, re-rendering of past inline searches.
AI briefs
When a brief is generated for a product:
- The brief text (Opportunity / Risks / Positioning / Sourcing).
- Score reasoning text.
- The (asin, marketplace) key.
- Cache TTL (48 hours).
Table: research_ai_briefs.
Note: Briefs are shared across users — same (asin, marketplace) key, same brief, no matter which user triggered the generation. This isn’t user-personal data; it’s product-research-level data.
Exports
For each export job:
- Export UUID and metadata (search UUID, format, status, timestamps, file size).
- The generated file — stored on disk, expires after 1 hour.
- Credits charged.
Table: research_exports (metadata) + filesystem (the actual PDF/CSV).
Why: Status polling, file delivery, credit tracking. The file is automatically purged after 1 hour to limit storage retention.
Credit ledger
For every billable action:
- The credit charge (amount, type, search UUID).
- Refunds (when applicable, with reason).
- Timestamps.
Tables: user_credits (current balance), credit_events (audit log).
Why: Billing reconciliation, dispute resolution, support.
What we don’t store
- Your Amazon Seller account credentials. All Amazon data flows through hilal-sp-api which holds the auth on Hilal’s side; Product Research never sees a per-user Amazon token.
- Your payment card details. Hilal Software’s billing uses a third-party processor; Product Research never sees the card.
- Browser session data, cookies, IP addresses. The API records JWT-issued user/org IDs, not browser fingerprints.
- Backup copies of deleted data. Once you trigger a hard-delete, the data is genuinely gone; we don’t keep “in case you change your mind” backups beyond the soft-delete window.
How long things are kept
Most data is kept for the lifetime of your account, with these exceptions:
- Exports — 1 hour after generation, then deleted.
- AI briefs — 48 hours, then regenerated on next access.
- Soft-deleted items — about 30 days before hard-deletion, in case you change your mind.
→ Retention & policies for the full retention schedule.
How to delete
→ Deleting your research data.