Our methodology
Trust-Vault was born from a simple observation: the AI tools market is exploding, but it is difficult to know who to trust. Our mission is to bring transparency, objectivity and rigor to the evaluation of AI solutions.
Every tool listed on Trust-Vault goes through a structured evaluation process based on measurable, public criteria. This page explains exactly how we calculate the Trust Score, how we verify reviews, and what guarantees of independence we provide.
The Trust Score: 0 to 100
The Trust Score is our central trust indicator. It synthesizes four essential dimensions, each scored out of 25 points, for a total of 100.
Privacy
25 pointsRespecting user privacy is a fundamental criterion of trust.
- GDPR / RGPD compliance: data protection policy, legal basis, designated DPO
- Data handling: server location, encryption at rest and in transit, retention period
- Privacy policy: clarity, accessibility, recent update
- User rights: effective access, rectification, deletion, portability
- Third-party sharing: transparency about subcontractors and non-EU transfers
Accuracy
25 pointsA tool must deliver on its promises: reliable features, quality output, satisfied users.
- Feature reliability: do the advertised features work as expected?
- Output quality: precision, consistency and relevance of results
- User satisfaction: weighted average rating from verified reviews
- Availability: uptime, response time, incident management
- Customer support: responsiveness, available channels, quality of responses
Security
25 pointsTechnical security protects users against data breaches, vulnerabilities and abuse.
- Certifications: SOC 2 Type II, ISO 27001, HIPAA where applicable
- Encryption: TLS 1.2+ in transit, AES-256 at rest minimum
- Protection measures: multi-factor authentication, access control, logging
- Bug bounty program or third-party audits: existence and scope
- Incident history: known breaches, response time, post-incident communication
Transparency
25 pointsA transparent vendor inspires trust: clear pricing, complete documentation, open communication.
- Pricing clarity: displayed price, no hidden fees, explicit cancellation terms
- Documentation: user guides, API docs, public changelog
- Open source: accessible source code, accepted contributions
- Communication: blog, regular updates, public roadmap
- Terms of service: readable, balanced, dated updates
Calculation formula
Each pillar is evaluated out of 25 points. The Trust Score is the direct sum of the four pillars:
Trust Score = Privacy (25) + Accuracy (25) + Security (25) + Transparency (25)
The final score is an integer between 0 and 100. It is automatically recalculated every week.
Score interpretation
4-step evaluation process
Every tool follows a rigorous process before receiving its Trust Score.
Submission or discovery
Tools are either submitted by their vendors through our form or identified by our technology watch team. In both cases, the tool enters our evaluation queue.
Expert analysis (15+ criteria)
Our team examines the tool against more than 15 objective criteria across the 4 Trust Score pillars. We test the tool, analyze its documentation, verify its certifications and evaluate its business model.
Community review verification
Reviews from our community of verified users complement the analysis. The accuracy score directly integrates real user satisfaction, weighted by review volume.
Continuous monitoring and updates
The Trust Score is not static. We continuously monitor each tool's evolution: new certifications, security incidents, policy changes, user feedback. The score is recalculated every week.
Our independence
The integrity of our evaluations is the cornerstone of Trust-Vault. Here are our commitments:
No pay-for-ranking
No vendor can pay to improve their Trust Score or position in our rankings. The score is algorithmically calculated from objective criteria. Period.
Affiliate disclosure
Trust-Vault uses affiliate links to fund the platform. These links are clearly identified. They have no influence on the Trust Score, ranking or content of our evaluations. A tool without an affiliate link is treated exactly the same way.
Editorial independence
Our editorial team is independent from the commercial team. Reviews, guides and comparisons are written without external influence. We never publish sponsored content disguised as objective evaluation.
Review verification
User reviews are an essential component of the Trust Score. We verify them rigorously:
Mandatory authentication
Only registered and verified users can publish a review. Each account is linked to a confirmed email address.
Anti-fraud detection
Our system automatically detects suspicious behavior: multiple reviews from the same IP, mass account creation, abnormal text patterns.
Human moderation
Every review goes through human moderation before publication. Obviously fake, defamatory or promotional reviews are rejected.
Community reporting
Users can flag a suspicious review. Our team examines each report within 48 hours.
Logarithmic weighting
Review volume uses a logarithmic scale: going from 10 to 100 reviews has a significant impact, but going from 1,000 to 10,000 adds almost nothing. This prevents volume manipulation.
Update frequency
The AI market evolves rapidly. Our methodology adapts continuously:
Weekly recalculation
Each Trust Score is automatically recalculated every week to integrate new data: new reviews, policy changes, certification updates.
Continuous monitoring
Our monitoring system tracks security incidents, terms of service changes and major vendor announcements. A significant event triggers an immediate recalculation.
Proactive alerts
In case of a significant Trust Score drop for a tool, we alert users who have added it to their favorites. Transparency goes both ways.
Methodology evolution
Our methodology itself is a living document. We adjust it regularly based on community feedback, regulatory changes and new threats. Every change is publicly documented.
Questions about our methodology?
We believe in total transparency. If you have questions, suggestions or concerns about our evaluation process, our team is here to listen.