Key concepts
These are the words and ideas you’ll see again and again. Once you know them, the rest of the section is fast to navigate.
SellerScore
A composite 0–100 score for each product. Combines six sub-scores by configurable weights. The single number you sort and decide by.
Default weights:
| Sub-score | Default weight |
|---|---|
| Demand | 25% |
| Margin | 20% |
| Competition | 20% |
| Trend | 15% |
| Seasonality | 10% |
| Cross-platform | 10% |
Each sub-score is also visible on the row, so you can see which of the six is dragging the composite up or down. → SellerScore explained
Sub-scores (the six)
| Sub-score | Measures | Source |
|---|---|---|
| Demand | How many people want it | SFR (Brand Analytics) where available; BSR + Google Trends fallback |
| Margin | How profitable each sale is | Sell price − COGS estimate − FBA fee − referral fee |
| Competition | How crowded the niche is | Review counts, avg ratings, listing density |
| Trend | 12-month direction | Google Trends 12-month delta |
| Seasonality | How spiky vs. smooth | Google Trends monthly variance |
| Cross-platform | Pricing leverage from other platforms | AliExpress / TikTok / Google Shopping (Coming soon — placeholder today) |
Search modes
- Standard — single-shot search, returns top 10 (or 50+ on expand). The Run-2 mode you have today.
- Deep / AutoSearch — iterative, multi-round refinement until convergence.→ AutoSearchComing soon
Entry modes
- Guided — form-based search (categories, fulfillment, price band, etc.).
- Chat — free-text natural-language search; AI extracts the parameters.
Both end in the same ResearchSearch snapshot in your history, with the same SellerScore.
Credits
Product Research uses a credit model. New users start with 3 free credits.
| Action | Credits |
|---|---|
| Standard search | 1 |
| Search expand (50+ results) | 2 |
| Product detail (first view) | 1 |
| Product detail (repeat views) | 0 |
| Chat (no triggered search) | 0 |
| Chat with triggered search | 1 (for the search) |
| Export (PDF or CSV) | 1 |
| History re-fetch | 0 |
| History re-run | 1 (it’s a fresh search) |
Credits are bundled into your Hilal Software plan. → Plans & credits
Marketplace
The Amazon regional storefront a search runs against. US, UK, DE, FR, IT, ES, CA, AU. Product Research uses the marketplaces you’ve configured for the rest of Hilal Software (Settings → Marketplaces).
Search snapshot
Every search produces a frozen snapshot — the products as they were at the moment of the search. Re-opening the search later shows the same numbers; that’s by design (you may have made decisions based on those numbers, and they shouldn’t drift). To get fresh data, re-run the search.
Watchlist
A list of products you’ve saved. Each watchlist row keeps a cached copy of the product’s metrics; you can refresh on demand if the cache is more than 30 minutes old.
AI brief
A Claude-generated paragraph per product covering opportunity, risks, positioning, and sourcing. Cached 48 hours and shared across users — meaning you don’t pay AI compute for a brief another seller already triggered for the same product.
Score reasoning
A separate Claude-generated explanation focused on why the product scored what it scored — useful when you need to explain a recommendation to a partner.
Data quality
Each search result carries a data_quality score (0–100) and a list of warnings when a sub-score had to fall back from the preferred data source. Trust low-quality results less.
Crawlee (Coming soon)
The microservice that’ll provide cross-platform pricing data (AliExpress, TikTok Shop, Google Shopping). Stubbed in Run 2 — the cross-platform sub-score is a neutral placeholder until Crawlee ships in Run 3.