Verification playbook
Our testing process
We follow a verification-first playbook for every survey and GPT app: static pages for speed, edge Workers for validation, and a data moat of human-proofed logs that AI summaries cannot fake.
Static-first publishing
We pre-render every review, guide, and chart for speed, then hydrate live trust signals—hourly rates, scam alerts, and leaderboards—through edge fetches so content stays fresh without slowing the first paint.
Evidence intake & scrubbing
Cloudflare Workers accept uploads, strip EXIF metadata, fingerprint images with SHA-256, and push sanitized payout screenshots to R2 before any public display.
Data integrity in D1
Relational tables log sessions, payouts, and scam reports. We calculate rolling hourly rates and payout reliability inside D1, then cache leaderboards in KV for millisecond reads.
Dynamic SEO at the edge
HTMLRewriter injects up-to-date schema (AggregateRating, Dataset, ClaimReview) into static HTML so search engines see verified survey numbers, not stale snapshots.
Data pillars
Earnings logs
User_id, platform_id, activity type, minutes spent, and earnings fuel the Real Hourly Rate Index and disqualification matrix.
Proofs
R2 object keys, OCR matches, verification status, and hash fingerprints back the payout proof gallery and leaderboard eligibility.
Scam reports
Incident dates, loss amounts, support transcripts, and scam categories build collapse timelines and payout reliability scores.
Users & trust
Trust scores, badge tiers, and streak counters weight every contribution so data from verified earners carries more influence.
Safety & anti-fraud
- OCR cross-checks: claimed payout amounts must match text extracted from the receipt before we accept a proof.
- Duplicate defense: image hashes are compared against prior submissions to block re-uploads from Google Images or other sources.
- Role-based review: higher-trust community members form a tribunal to vote on edge cases after automated checks run.
- Privacy defaults: GPS and device metadata are stripped, and proofs are redacted to keep personal identifiers off the site.
Experience signals
Each review, guide, and comparison chart links back to a verifiable signal:
- Aggregate ratings built only from verified users and tribunal-cleared payout proofs.
- Time-to-cashout distributions from session logs so readers see the work required.
- Copy that answers payout friction, VPN rules, and ID checks to defend against AI fluff.
How to contribute
Log earnings
Record minutes spent and payouts per session to strengthen hourly rate modeling.
Upload proofs
Submit redacted receipts; Workers scrub metadata, run OCR, and reject mismatches before they hit the gallery.
Audit scams
File scam reports with dates and screenshots. Tiered reviewers validate them to keep the collapse timeline honest.