Analyze insolvency risk
Enter one CUI or a list of CUIs to see the estimated insolvency probability and risk category.
Promotional access for signed-in accounts until 31 December 2026. From 1 January 2027, an active access package is required.
Sign in to use the insolvency prediction.
Predictive tool for commercial risk analysis
Estimated insolvency probability for more prudent B2B decisions
We estimate the statistical insolvency probability of a company using available financial and non-financial data (CAEN, employees, court cases, company age, number of directors, etc.), based on predictive models tested on prior-year data. The probability is shown as a percentage. If you see 14%, that means an estimated 14% risk, not an opaque score. To make it easier to interpret, the percentage is also paired with a risk category: low, moderate, medium, or high.
Built on in-house research, not on an oversimplified formula
The model was developed for practical use in commercial risk analysis, with an emphasis on stable results over time and prudent interpretation. The selected model showed much higher accuracy than Altman Z-score style approaches.
Cross-functional team
A team of 5 engineers, including 3 specialized in AI and machine learning, working alongside an accountant experienced in interpreting financial data.
Over 30 models analyzed
We analyzed more than 30 promising ML models and selected the combination that delivered the most stable results in our simulations.
10-year simulations
The simulations covered a 10-year interval to observe model behavior across different economic contexts.
Self-funded investment
The predictive model was developed over 6 months through a substantial self-funded investment focused on accuracy and precision. This places it above many projects built through European funding.
How it works
The service uses both financial data and relevant non-financial signals to generate an estimated insolvency probability that directly represents the predictive risk.
Relevant company data
We start from financial data, derived indicators, and other relevant non-financial signals about the company, shareholders, directors, court cases, and related companies.
Predictive model
We apply a combination of statistical and machine learning models selected through testing on historical data.
Estimate shown as probability
The result is shown directly as an estimated insolvency probability together with a statistical risk class. Unlike a score, the probability shows how large the risk is in concrete terms, not just that one company is riskier than another.
Decision support
The risk can be used as an internal signal for additional checks, monitoring, and prioritization.
How you can evaluate in practice whether understanding the risk helps you
Check companies that already caused you problems
You can test the companies that caused delays, losses, disputes, or insolvencies in the current year and see whether there were signals that deserved closer analysis.
Analyze your portfolio of clients and suppliers
Upload your partner list and see where additional checks, a more prudent commercial limit, or closer monitoring would have been justified.
Test past scenarios
For historical evaluations, the prediction is made with a model trained only on data available up to that year, without using information from the year we want to predict. This helps you see whether losses could have been avoided.
Risk should not be ignored, but prioritized
In B2B relationships, treating every partner the same can be costly. Predictive risk does not decide for you, but it shows where it is worth investing time in additional checks.
A practical advantage of the model is that it allows risk management through more gradual adjustments, with less dramatic sales drops than in situations where the reaction comes too late.
- you can request advance payment or guarantees where the risk is high;
- you can reduce commercial limits for partners that require caution;
- you can monitor more closely the companies that fall into risk areas;
- you can avoid discovering too late signals that were already present in the data.
Interpretation guidelines
- correct interpretation should be made together with other commercial, legal, and financial information, and decisions remain the user's responsibility;
- a well-calibrated probability is more useful than an opaque score: for companies estimated at 40%, we expect about 40% of them to enter insolvency, while a score only shows the ranking of risk, not its magnitude;
- the risk shown for 2026 is calculated as of January 1, 2026 and should be interpreted as valid for the full year, based on the information that could be used at that moment;
- the calculation does not include information that appeared after January 1, 2026, such as insolvencies opened later or tax debts published later; for that reason, a company that entered insolvency during 2026 may still appear with a subunitary risk, exactly as it could have been predicted on January 1, 2026;
- the evolution of risk over time can be just as important as its point-in-time level;
- for decisions with high impact, an additional verification is recommended before taking action.
What predictive analysis provides
Probability, risk category, and trend over time provide additional guidance when reviewing a company or portfolio.
Estimated insolvency probability
A probabilistic estimate of insolvency risk for a defined time horizon.
Statistical risk class from low to high
A quick classification that helps triage companies and prioritize checks.
Results for one CUI or for a list of CUIs
You can quickly analyze one company or several companies in a portfolio.
CSV export
Export for portfolio reviews and analysis.
Results are generated automatically based on the available data and provide a probabilistic estimate, not a certainty. They should be interpreted together with internal procedures and other relevant sources about the company.
The predictive model did not estimate insolvency risks above 80%, indicating a limitation in detecting extreme risks that may be influenced by causes not included in the training dataset.
The methodology uses a combination of statistical and machine learning models selected after internal testing. The risk is calculated automatically and deterministically based on the available data for the specified CUI.
The estimated risk is intended for use as an internal decision-support tool. It should not be used as the sole basis for lending decisions, investments, or other decisions with significant legal or financial impact.
The result should be reviewed together with the other available company information.
Sign in required for activation
Sign in with an eligible account to use the insolvency prediction.
The indicators presented are statistical and predictive tools for internal analysis, do not constitute a regulated credit rating, and must not be used for regulated purposes.