Scoring
Methodology
How SignalScore assesses Swiss company creditworthiness using publicly available data and AI analysis.
7
Dimensions
12
Data Sources
Claude
AI-Powered
Section 01
How It Works
OSINT-based credit assessment designed for Swiss companies
Collect
12 public data sources queried in parallel — official registries, web presence, and public media.
Analyze
Claude (Anthropic) reviews the raw data and writes a transparent, source-cited explanation of each of the 7 dimensions.
Score
A deterministic formula scores the 7 weighted dimensions 0-100 from the raw data and combines them into a composite (0-100) and letter grade (A-F) — an indicative assessment.
The final score represents an indicative assessment, not a certified credit rating. See our Disclaimer for important limitations.
Section 02
The 7 Scoring Dimensions
Each dimension weighted by empirical predictive power
Weight Distribution
Registry & Legal Stability
- Company age, legal form (GmbH, AG, etc.), and active/cancelled status
- Share capital amount and registered address
- SOGC publication history — bankruptcies, liquidations, capital changes, officer mutations
- Art. 731b organizational deficiency signals
Digital Presence & Infrastructure
- Website existence and content quality (crawled via Firecrawl)
- Domain registration age (via RDAP lookup)
- HTTP security headers (HSTS, CSP, X-Frame-Options)
- Server technology detection and email infrastructure (MX records)
Workforce & Growth
- Estimated employee count and company size (from public search results)
- LinkedIn company page presence and follower count
- Active job postings on jobs.ch, Indeed, and LinkedIn Jobs
Reputation & Public Perception
- Google Maps rating and review count
- Business listing status (operational, temporarily closed, etc.)
- News media coverage — adverse-media screening in any language (bankruptcy, debt enforcement, layoffs, litigation, fraud). Negative coverage lowers the score and is weighted more heavily than positive coverage; article volume alone is never treated as positive.
Officer & Governance Analysis
- Officer and director changes visible in SOGC publications over time
- Frequency and pattern of board-level mutations
- Presence of audit company and governance structure
- Purpose statement changes and company name history
Analysis is based on the target company's own SOGC publications. Cross-company officer network analysis is not currently available.
Business Listing Verification
- Google Maps business listing existence and status
- Registered address matches a real business location
- Business type classification and opening hours
Verification is based on Google Maps listing data. No Street View imagery or physical site inspection is performed.
Temporal & Change Velocity
- SOGC publication frequency and timing patterns
- Timeline of officer changes, capital modifications, and legal mutations
- Clustering of negative signals within short time periods
Temporal analysis is primarily based on the SOGC publication timeline. Other data sources (website, LinkedIn, Maps) provide a current snapshot only.
Section 03
Scoring Scale
Six-tier grading from Very Low Risk to Very High Risk
Very Low Risk
Score 85-100
Low Risk
Score 70-84
Moderate Risk
Score 55-69
Elevated Risk
Score 40-54
High Risk
Score 25-39
Very High Risk
Score 0-24
The composite score is calculated as a weighted sum of all 7 dimension scores (each 0-100), then mapped to the letter grade above. Individual dimension scores are visible in every report.
Section 04
Data Sources
12 public data sources queried in parallel for every credit check
Zefix
Swiss Central Business Name Index — official company registry data via REST API with SPARQL fallback
SHAB/SOGC
Swiss Official Gazette of Commerce — historical legal publications extracted from Zefix records
Cantonal Register
Official cantonal commercial register — exact registration date extracted via Firecrawl from JSF-rendered pages
UID Register
Business Identification Register — VAT status, NOGA industry code, legal form, address (via LINDAS SPARQL)
Company Websites
Discovered via Google search, content crawled via Firecrawl (up to 15,000 characters)
Estimated company data from public Google search snippets (not direct API access)
Google Maps
Places API — business listing, rating, review count, status, address, opening hours
News & Media
News via SerpAPI Google News + GDELT (multilingual, any language), including a dedicated adverse-media pass for bankruptcy, debt enforcement, layoffs, litigation and fraud. Negative coverage lowers the rating; it is routed to the matching risk dimension and weighted more heavily than positive coverage.
DNS/RDAP
Domain registration age, HTTP security headers, server technology, and email infrastructure (MX records)
Job Portals
Active job postings on jobs.ch, Indeed, and LinkedIn Jobs — hiring activity signals
Wayback Machine
Internet Archive snapshots — oldest snapshot date indicates digital presence longevity
SwissReg
Swiss Federal Institute of Intellectual Property — registered trademarks and patents
All data is sourced from publicly accessible registries and websites. No non-public, proprietary, or confidential data is used. LinkedIn data is estimated from Google search snippets, not from direct API access. See our Privacy Policy for details on data processing.
Section 05
Academic Foundation
Dimension weights informed by peer-reviewed research and commercial methodologies
Creditreform Bonitätsindex
Dominant DACH credit scoring methodology. 11 weighted factors including company age (4%), legal form (4%), employee count (4%), and industry risk (6%). Our dimension weights are derived by scaling Creditreform's accessible factors to 100%.
BFS UDEMO — Swiss Business Demography
Swiss Federal Statistical Office survival curves: 84%/72%/64%/51% at years 1/2/3/5. Calibrates our age scoring tiers. 11,506 bankruptcies in 2024 by sector informs industry risk signals.
Altman Omega Score (JSBM)
Journal of Small Business Management, Vol 61(6). Extended Z-Score for SMEs with LASSO-selected non-financial variables: management changes and employee tenure. AUC 88%. Validates our officer stability and workforce dimensions.
Ferretti et al. — Bayesian SME Default
Journal of Management and Governance. CEO tenure coefficient -4.18 (p<0.0001) — strongest governance predictor. Non-default firms: avg tenure 26.5y vs. defaulting firms 5.3y. Anchors our Officer & Network dimension.
Crosato, Domenech & Liberati — Website Indicators
Economics Letters, Vol 204. 50 binary website indicators outperformed traditional offline data (employees, debt, profit) for SME default prediction. Validates our Digital Presence dimension with 15% weight.
Berg, Burg, Gombovic & Puri — Digital Footprints
Review of Financial Studies, 33(7). 250,000+ observations: digital footprint variables equal or exceed credit bureau scores for default prediction. Confirms that online signals carry genuine predictive power.
Shumway — Hazard Models for Bankruptcy
Journal of Business, 74(1). Pioneered discrete-time hazard models showing monotonically decreasing failure rate with age (0.031 at founding, ~0.015 by month 40). Underlies our temporal scoring logic.
Ciampi — Corporate Governance & Default
Journal of Business Research, 68(5). Management-related variables (CEO duality, ownership concentration) significantly improve small enterprise default prediction. Validates governance as a scoring dimension.
Section 06
Swiss Context
Record bankruptcy levels make creditworthiness assessment critical
0
Bankruptcies in 2024
Record high, +15% YoY (BFS)
0%
5-Year Survival Rate
BFS UDEMO new company cohorts
0.8%
Annual Default Rate
Swiss overall (2024, elevated)
Age Distribution of Failures
KOF/ETH Zurich analysis reveals that bankruptcies are disproportionately concentrated in the 4-10 year age group, not the youngest firms. This "liability of adolescence" pattern means simple company age is insufficient — our model uses age as one factor alongside governance and publication signals.
Sector Concentration
Share of Swiss bankruptcies by sector (2024)
Section 07
Calibration & Transparency
Honest methodology builds trust
The current dimension weights are informed by academic literature — starting points, not empirically calibrated against Swiss default outcomes yet. We are transparent about this because we believe honest methodology builds trust.
Current Limitations
Officer analysis limited to target company — no cross-company network analysis yet
Temporal analysis relies on SOGC timelines; other sources are current snapshots only
LinkedIn data estimated from search snippets — accuracy varies
Geospatial verification uses Google Maps listings, not physical inspection
Planned Improvements
MVP Validation
Compare against known SHAB bankruptcy outcomes
Back-testing
Retrospective analysis of 2023-2024 failures
Historical Tracking
Enable real temporal comparison across all sources
Weight Optimization
Find most discriminating weight configuration
Important Limitations
- SignalScore is a non-financial signal aggregator using only publicly available data — no balance sheets, no payment history
- Dimension weights are derived from Creditreform factor scaling and peer-reviewed research (Crosato 2021, Ferretti 2021, BFS UDEMO)
- Age scoring calibrated to Swiss BFS survival curves; industry risk from actual 2024 Swiss bankruptcy distribution by NOGA sector
- Not a substitute for professional credit bureaus (Creditreform, D&B, CRIF) which have access to financial statements and payment data
- Must not be the sole basis for credit, employment, or legal decisions
- AI generates narrative explanations only — all scores are computed by a deterministic formula engine
Methodology informed by peer-reviewed academic research and Swiss corporate failure data.
© 2026 SignalScore by Predivo GmbH. All rights reserved.
Swiss-made