Built like infrastructure, not like a demo.
Rankow runs unattended on your sites, so the platform has to be safe by default: isolated tenants, encrypted credentials, reversible publishes, and spending that stops before it exceeds scope.
Integrations
Official API access
SERP & keyword data API
Keyword, SERP, and competitor data come from official search-market APIs, metered per request and written to the run ledger.
Google Search Console (OAuth)
Performance data is read from your own verified property via OAuth — impressions, clicks, and positions as Google reports them.
Shopify Admin API (GraphQL)
Publishing and rollback run through the official Admin GraphQL API with scoped credentials and idempotent writes.
Platform blocks
Six parts of a safe-by-default platform.
Every tenant gets the same enforcement: isolation, encryption, permissions, reversible publishing, reliable runs, and cost guardrails.
Tenant isolation
Your sites, keywords, and articles live in rows that only your tenant can reach.
- Postgres row-level security is enforced, not just filtered in application code.
- Data access runs through tenant-scoped policies verified in automated checks.
- No cross-tenant queries by design; the guard also blocks empty-context reads.
Credentials & authentication
Secrets are encrypted at rest; sessions and keys follow platform hardening basics.
- Site credentials are encrypted with AES-GCM before they touch the database.
- Passwords are hashed with argon2id; sessions use httpOnly, SameSite cookies.
- API keys are stored only as SHA-256 digests — the plaintext never persists.
Roles & permissions
Four roles with a minimal permission matrix, enforced server-side.
- Owner, admin, editor, and viewer cover the team without sharing everything.
- Nine permission points — from tenant management to publishing — are mapped centrally, not scattered in business code.
- Billing and publishing actions require the explicit roles that own them.
Publishing safety
Every publish is reversible, and a failure never fakes success.
- Writes are idempotent — a retried publish cannot duplicate an article.
- Every publish stores a pre-change snapshot; a >3 position drop can be rolled back with URLs unchanged.
- SEO metafields and structured data (JSON-LD) are written as part of the publish, and sitemaps stay in sync.
Run reliability
Scheduled runs survive crashes and never bill twice for the same work.
- Runs hold a site lock with heartbeat renewal; a crashed worker's stale jobs are recovered automatically.
- Retries reuse completed stages from the parent run — no duplicated generation, no duplicated cost.
- Cron schedules respect your timezone, skip duplicates within a site, and catch up missed runs once.
Cost guardrails
You see what every run costs, and a run stops before it overspends.
- Keyword, SERP, Search Console, LLM, and image usage are metered per run into a ledger.
- A budget gate checks the ledger before spending; exceeding the agreed scope stops the run, not your bill.
- Quota errors return structured codes with limit, usage, and reset date — no silent overage.
Next step
Want the platform detail first-hand?
A pilot cycle runs the full loop on one site and one section — including the ledger, the gate, and the 28-day read.
Reply within 48 hours · No slide deck