Transparency
How We Score the World
Every score is reproducible. Every methodology is public. Affiliate relationships are disclosed and never influence our ratings.
Key Takeaways
GemmWork earns affiliate commissions from some of the services we recommend. Our scoring is calculated independently using the GEMM framework — not based on those relationships. We do not accept placement fees, sponsored rankings, or advertising from any EOR provider or talent platform.
How GEMM Scores Are Calculated
Every country recommendation in GemmWork is generated by the GEMM scoring engine — a deterministic algorithm that calculates a weighted composite score across 6 variables. No human editorial judgment is applied to the rankings. The same inputs always produce the same outputs.
| Variable | Weight | Definition | Primary Sources |
|---|---|---|---|
| Cost Efficiency | 25% | Salary savings vs US equivalent role. Formula: (US Salary − (Local Salary + EOR Fee)) / US Salary | Near, South, Howdy (2026) |
| Cultural Proximity | 20% | English proficiency (EF EPI), timezone overlap vs EST, US business culture alignment | EF EPI 2025, GemmWork analysis |
| Compliance Stickiness | 15% | Ease of exit — severance law, EOR duration limits, misclassification penalty | OECD Employment Protection Index, local labor law |
| AI Workflows IQ | 15% | AI tool adoption rate + Agentic Workflow Readiness (AWR). Formula: TP×0.4 + AWR×0.6 | GitHub, Kaggle, Coursera (2026) |
| PE Risk (inverse) | 15% | Permanent establishment risk under OECD 2025 rules. Higher score = lower risk | OECD Model Tax Convention 2025, EY, KPMG |
| Data Risk (inverse) | 10% | Data privacy law exposure. Higher score = lower risk | GDPR, DPDP Act, LGPD, Ley 1581 |
How Weights Adjust Based on Your Inputs
The default weights above apply to most company profiles. GemmWork's engine automatically adjusts weights based on specific inputs:
- GDPR-sensitive data: Data Risk weight increases to 20%. Recommended for companies handling EU personal data.
- Low risk tolerance: PE Risk weight increases to 25%. Triggered when risk_tolerance = “low”.
- US timezone required: Cultural Proximity weight increases to 30%. Triggered when us_timezone_required = true.
- ML/AI roles: AI Workflows IQ weight increases to 25%. Triggered for ml_engineer and data_engineer roles.
These adjustments are applied automatically in the GEMM Navigator. The adjusted weights are shown in your results PDF report.
Our Affiliate Relationships
GemmWork participates in affiliate programs. When you sign up for a service through our links, we may earn a commission — at no additional cost to you. Below is a complete list of our current affiliate relationships.
| Partner | Program | Commission | Our Assessment |
|---|---|---|---|
| Deel | PartnerStack | $500 (lead) + $1,000 (customer) | Independently scored via GEMM |
| Remote.com | Impact | 10–15% revenue share (12 months) | Independently scored via GEMM |
| Toptal | Direct | $2,000/client | Independently scored via GEMM |
| Oyster HR | Direct | Est. $300–500/client | Independently scored via GEMM |
| Firstbase.io | PartnerStack | 10% per incorporation | Independently scored via GEMM |
| Wise | Partnerize | ~$40–50/referral | Independently scored via GEMM |
| Payoneer | Direct | $25/referral | Independently scored via GEMM |
| NordVPN | Direct | 40–100% first month | Independently scored via GEMM |
| DocuSign | CJ Affiliate | $25–100/signup | Independently scored via GEMM |
All links to these services on gemmwork.io route through gemmwork.io/go/[partner], which records the click in our database before redirecting. This allows us to track which content drives conversions — data we use to negotiate better terms and improve our coverage. We never receive placement fees for rankings, and our GEMM scores are not visible to any of these companies before publication.
Editorial Independence
GemmWork's editorial process operates independently of its affiliate relationships. The following rules are enforced without exception:
- No placement fees. We do not accept payment to rank a service higher.
- No pre-publication review. Affiliate partners do not see our scores or analysis before publication.
- Score immutability. GEMM scores are generated algorithmically. No manual override is permitted.
- Parallel analysis.Every affiliate partner's official claims are presented alongside GemmWork's independent analysis on all comparison pages.
- Negative findings published. If our analysis finds a risk or limitation, it is published regardless of affiliate status. See our Brazil compliance stickiness rating (High) as an example.
If you believe a score is incorrect, contact us. We review and update scores when presented with credible evidence.
Data Sources and Update Frequency
| Data Type | Primary Sources | Update Frequency |
|---|---|---|
| Salary data | Near, South, Howdy, Mismo (2026) | Annual |
| EOR pricing | Direct verification with providers | Based on August 2025 data |
| Employer tax rates | Official government sources, EY, KPMG | Based on August 2025 data |
| Severance / exit costs | Local labor law, EOR provider documentation | Based on August 2025 data |
| PE risk rules | OECD Model Tax Convention, EY, KPMG, Ogletree | Within 48hrs of regulatory change |
| Data privacy laws | Official government sources, Garrigues | Within 48hrs of regulatory change |
| AI talent metrics | GitHub, Kaggle, Coursera, HackerRank | Semi-annual |
| EF EPI (English) | EF Education First | Annual (EF publishes October) |
When a material regulatory change occurs — such as the OECD 2025 Model Tax Convention update — GemmWork updates affected content within 48 hours and increments the dateModified timestamp on all relevant pages. The JSON-LD schema on every page reflects the actual date of last meaningful update, not a rolling timestamp.
Contact
For corrections, data disputes, or affiliate partnership inquiries:
For score correction requests, please include:
- The specific country or mode score you believe is incorrect
- The source data you are citing
- Your proposed correction
GemmWork reviews all correction requests within 5 business days.