OpenData / Features / API access
The store database as an API
By the OpenData team · 15 Nov 2026
Files are for campaigns; APIs are for systems. The same OpenData index — 1,504,744 verified stores identified across 23,984,359 scanned domains — is queryable over HTTPS — GET /v1/stores with filters for platform, technology, app, payment gateway, country and has-email — through RapidAPI or the Apify actor, with a free tier large enough to build against.
OpenData index — 23,984,359 domains scanned · 1,504,744 verified stores · snapshot 15 Nov 2026
One endpoint, the full record
The API has a deliberately small surface: GET /v1/stores, eight filter parameters, cursor pagination. Each returned row is the same 12-field record the paid files carry — domain, store name, platform, country, currency, technologies, apps, payment gateways, public business email where the store publishes one, store-published phone, source URL and last-observed date. The full field list is in the column spec.
Because the API and the files read the same index, a segment you validate through the API is the segment you would get in the file — no bait-and-switch between the demo and the deliverable. The index behind both holds 1,504,744 verified stores out of 23,984,359 domains scanned (snapshot 15 Nov 2026).
Route 1: RapidAPI
The OpenData listing on RapidAPI is the managed route — keys, quotas and billing handled by RapidAPI. The Basic plan is free: 1,000 requests a month, with a 50-row cap per request for API callers; Pro is $20/month for 500,000 requests. The row cap is deliberate: the free tier is for building and validating, not for draining the database.
Full request/response documentation lives on the developers page.
Route 2: the Apify actor
For batch pulls inside the Apify ecosystem, the OpenData actor runs the same filters as a scrape job at $4 per 1,000 rows. It suits scheduled refreshes — a recurring segment pull into a dataset — better than interactive querying. Both routes return identical records.
Route 3: the MCP server
For AI agents and MCP clients, OpenData also runs a live MCP server (Model Context Protocol, Streamable HTTP) at https://opendata-mcp.superblackhero8.workers.dev/mcp. It needs no key on the public tier and exposes three tools: count_stores (counts by platform, country, tech, app, gateway and email availability), search_stores (up to 25 matching stores per call) and get_store (the full public record for one domain).
The public tier shows domains, platforms, technologies, apps and gateways in full; email addresses are reported as available or not, never printed — the addresses themselves ship in the paid files and the RapidAPI plan. The MCP server answers from the 4 Nov 2026 dataset snapshot (814,516 stores), while the on-site checker runs on the newer snapshot. Connection details and a client configuration example are on the developers page.
API or file? A straight decision rule
Choose the file when your ICP is stable: it is the whole segment, billed per row and yours to keep, refreshed monthly. Choose the API when the query changes — a tool that looks stores up one by one, a CRM enrichment step, or an AI agent assembling a fresh slice per task. Many buyers do both: the file for the core market, the API for the long tail. The ICP filtering page shows how the same filters work across all three surfaces.
Either way, start free: the RapidAPI Basic tier, the store checker (30 unlocks a month) and the industry samples all read the same live index, so what you test is what you get.
Questions, answered straight
What does the API cost?
The free Basic plan allows 1,000 requests per month with up to 50 rows per request. The Pro plan is $20/month for 500,000 requests. Both are on the OpenData listing on RapidAPI; the underlying records are identical.
What does one API result contain?
Each result is the full 12-field record: domain, store name, platform, country, currency, technologies, apps, payment gateways, public business email where published, store-published phone where published, source URL and last-observed date.
Which filters does the API support?
GET /v1/stores accepts platform, tech, app, gateway, country and has_email, with cursor pagination so large pulls never return duplicates. The same filters drive the catalog and the paid files.
When is a file better than the API?
Pull the rows. A per-row pull (full records $4 per 1,000 rows, platform records $1 per 1,000 rows) covers the complete segment and exports immediately as CSV or Excel; the data is refreshed monthly. The API earns its keep when filters change often or software does the querying.
Can AI agents query the database directly?
Yes — the API is the natural interface for agents: the query is a URL, the answer is JSON records. Agent workflows typically pull small, tightly filtered slices (a platform, a country, has_email) rather than bulk exports.