How credits work
Prepaid, transparent, and always priceable before you commit.
The model
- One prepaid credit wallet covers the Data API, the Scraping API and the MCP connector.
- A Data API data row costs 5 credits; catalog/discovery calls cost 1 credit.
- Scraping products are billed per item (per product page, per keyword, per shop) — the rate card lists each one.
- Estimates are always free and exact — you see rows and credits before anything is charged.
- At zero balance requests are refused, never invoiced. No surprise bills.
A worked example
Say you want every SKU row for Facial Serum in Indonesia for one month. First, the free estimate:
POST /v1/exports {"country":"ID","category_3":["Facial Serum"],
"date_from":"2026-05","date_to":"2026-05","estimate":true}
→ {"estimate": {"rows": 289621, "credits": 1448105, "usd": 1810.13}, "charged": 0}289,621 rows × 5 credits = 1,448,105 credits. On the Tier 3 pack that's about $1,608; pay-as-you-go about $2,172. If the price is wrong for your use case, you narrow the scope — fewer months, a deeper category — and estimate again. Nothing is charged until you submit without the estimate flag.
Guard rails
The Python client adds a hard spend ceiling: max_credits on a submit refuses to run if the scope grew since you estimated. Failed scraping jobs are refunded server-side; free client-side validation catches malformed items before anything is spent.