Solventa cuts fraudulent transactions by 25% and implements SEON’s ML to boost accuracy by 15%

Solventa cuts fraudulent transactions by 25% and implements SEON’s ML to boost accuracy by 15%

Solventa wanted to reduce its loan defaults arising from synthetic and fake IDs, while increasing its customer base.

With SEON it gathered alternative data from social and digital on potential customers, assessed their legitimacy, appropriate loan size and the likelihood they would make repayments.

This led to a 25% drop in fraudulent transactions, and with SEON’s machine learning Solventa was able to improve the accuracy of its fraud scoring system by a further 15%.

About Solventa:

Solventa is the leading Fintech provider of immediate loans in Colombia, Peru and Mexico. It qualifies individuals and SMEs in minutes, facilitating greater financial inclusion across those countries


For Solventa and other companies in the consumer lending sector, the impact of fraud is widespread. Fraudsters are constantly attempting to steal IDs and impersonate customers to secure a loan they won’t repay.

Therefore, Solventa needed to identify potential fraudsters, without adding unnecessary onboarding friction which could slow down its rapid loan disbursements – a key factor in its service. 

Solventa also knew there was a large market for under/unbanked individuals which alternative data would be able to provide a background for, along with data for determining risk/ratio of likelihood to default on payments.

Lu Vera, partner at Solventa, explains:“Our mission is to provide credit to all legitimate customers we are able to, and we aim to do this fast – often in under a few minutes. So we have to make sure we are doing our due diligence and doing it as quickly as possible. We were mostly using manual checks, such as cumbersome ID validations in public databases, phone calls, and security profiles based on highly qualitative data, but we were unsatisfied with the time and resources required.” 


By implementing SEON’s rapid real-time API module Solventa could check its customer’s social and digital footprint with only their phone, email, or IP address. With it, Solventa gained insight into its customer’s financial stability and therefore risk of default or fraud. 

Solventa also applied SEON’s in-depth custom risk scoring system to its customers. Immediately opting to accept them for a loan, or to be reviewed if the likelihood of fraud or default was high.

Lu shares: “We are currently implementing SEON into our Credit scoring flow as an added ID and security step. Other key variables, especially those associated with email registration and phone history dates, serve as credit predictors as well.”

By using alternative credit data from SEON’s email and phone lookup, Solventa can see whether a loan applicant has any of the 90+ social and digital accounts, such as LinkedIn, Facebook or AirBnB. This gives a strong indication of whether they are either a fraudster who is using fake details, or a legitimate person who has a lack financial means to repay the loan.

For example, while a fraudster is likely to have little to no digital footprint, accounts such as LinkedIn can be checked to assess a real applicant’s employment status and ability to repay a loan. Meaning Solventa can use this data to evaluate the size of the loan they are willing to offer, 

Solventa further benefited from SEON’s end-to-end capabilities and used SEON’s Machine Learning (ML) with the detailed labeling API to feedback data to the system. SEON’s whitebox ML then suggested new fraud prevention rules, based on locations, device fingerprinting and social signals.

Reduce Fraudulent Transactions by 25% & Decrease Manual Review Times

Allow rapid decision making for a frictionless onboarding experience. Leaving your team more time to help your business grow.

Speak with an Expert


SEON’s support resulted in a 25% drop in fraudulent transactions, which in turn lowered the cost of non-payment.

Furthermore, Solventa improved its risk-scoring models. When layered with SEON’s machine learning data, accuracy then improved by a further 15%. 

– 25% drop in fraudulent transactions

-15% increase in the accuracy of the machine learning models in detecting fraud

– Fewer defaulting loan repayments

– Expansion into underbanked markets 

– Savings on costly biometric and ID validations

– Increased productivity for manual reviews

In her final comments on the partnership with SEON, Lu mentioned the great collaboration with SEON’s customer support team:

“We are delighted with the timely turnarounds, quick responses, and effective problem-solving skills of all the analysts and staff working on Solventa’s account.”

For more information on Solventa, please visit or 

Share on social media