How We Work

Transparent by design.

Independent editorial judgement, cited sources, and transparency about our methods and limitations. The same principles across everything we publish.

Last updated: April 2026

What We Publish

Three Types of Content

Comparisons

Grade Scale

A letter grade system from A+ to F. Each tier reflects our overall assessment of a product's privacy and security posture.

A+ / A / A-

Excellent

Top-tier privacy and security. Strong encryption, verified no-logs or zero-knowledge architecture, independent audits, privacy-friendly jurisdiction, and open source code.

B+ / B / B-

Good

Solid privacy fundamentals with minor concerns. May have jurisdiction issues, corporate ownership questions, or limited audit history, but still demonstrates a genuine commitment to user privacy.

C+ / C / C-

Average

Functional product with room for improvement. May lack independent audits, have limited transparency, or collect more data than privacy-focused alternatives.

D+ / D

Below Average

Privacy is not a primary focus of the product. Data collection practices go beyond what is necessary for core functionality, in our editorial assessment.

F

Failing

In our editorial opinion, the product's data practices are fundamentally at odds with user privacy. This tier reflects our assessment of publicly available information.

Comparisons

Scoring Signals

No single signal determines a grade. The final assessment is a holistic editorial judgment based on all available evidence.

No-logs / zero-knowledge

Verified by independent audit

Unverified claims or known logging

Independent audits

Recent audit by reputable firm

No audits or audits older than 2 years

Jurisdiction

Privacy-friendly (e.g. Switzerland, BVI, Panama)

Five Eyes / Fourteen Eyes member

Open source

Client and/or server open source

Fully proprietary, no public code

Corporate ownership

Independent or privacy-focused parent

Owned by advertising / data company

Security track record

No breaches, or breaches handled transparently

Breaches concealed or poorly handled

Encryption

Strong, modern protocols (WireGuard, E2EE)

Weak or outdated protocols

Past incidents aren't permanent marks. Historical incidents (e.g. past breaches or logging events) are weighed against what the company has done since: a transparent disclosure followed by structural changes (RAM-only servers, new audits, jurisdiction changes) carries less weight than an unaddressed incident.

Comparisons

Category-Specific Criteria

What we assess for each product category.

VPNs

  • Encryption protocols and implementation
  • Logging policy and verifiable no-logs claims
  • Jurisdiction and legal obligations
  • Independent security audits
  • Ownership and corporate structure
  • Kill switch and leak protection
  • Speed and server network

Browsers

  • Telemetry defaults and opt-out options
  • Tracking protection and content blocking
  • Fingerprint resistance
  • Built-in ad blocking capabilities
  • Open source status and rendering engine
  • Update frequency and ownership

Email Providers

  • End-to-end encryption support
  • Zero-access encryption architecture
  • Jurisdiction and legal framework
  • Metadata handling and storage encryption
  • Open source status and data practices

Password Managers

  • Encryption standard and implementation
  • Zero-knowledge architecture
  • Independent security audits
  • Open source status and breach history
  • Platform support and data practices

Standards

Sources and Verification

Every claim traceable. Every source weighted. Every editorial decision ours.

Verifiable Sources

We cite our sources across all content types so you can verify our findings yourself. Research articles include numbered source lists. Comparison cards link to the evidence behind each assessment. Guides reference official documentation.

Source Reliability Tiers

T1
T2
T3

T1 Academic papers, regulatory filings, independent audit reports

T2 Established news coverage, expert analysis

T3 Company's own claims -- hedged with "claims to", "according to"

Evidence-Based, Not Lab-Tested

We do not test products in controlled lab environments. Our assessments are based on documented evidence, published research, and expert analysis of publicly available information.

AI-Assisted Research

We use AI tools (Claude by Anthropic) for research, fact-checking, and drafting assistance. Specific claims are verified against primary sources where possible. All methodology, editorial decisions, and final content are our own.

Policies

Our Commitments

Independence

Our grades are not influenced by affiliate relationships. Products with affiliate programmes and products without them receive the same editorial treatment.

Corrections and Right of Reply

If you believe we have made a factual error, we want to hear from you. Companies and individuals discussed in our content are welcome to respond. We will consider any evidence provided and update our content if the facts support it.

Updates

All content is periodically reviewed for accuracy. Research articles include changelogs. Comparisons are re-evaluated as products change. Grades may change over time in either direction.

All content on The Threat Model reflects our independent editorial analysis and opinion. It is published in good faith for informational purposes and is not professional security, legal, or financial advice. We make every effort to ensure accuracy but do not guarantee that all information is complete or current. See our terms of service for full details.