Trust in about 30 seconds

Trooly uses public and permissible digital footprints to understand and predict the trustworthiness of individuals and businesses. Using minimal and non-intrusive identity information as input, we return Instant Trust™ ratings which incorporate verification of the input identity, screening for undesirable past behaviors, and, using robust machine learning, prediction of likely future behaviors. Our patented platform and techniques ensure that our results are much more accurate, cost-effective, and free of bias than old school background checks, credit scores, and risk management tools. Trooly's Instant Trust™ ratings are immediately available via simple REST API's, a dedicated customer portal, or secure exchange of batch files. Trooly supports both one-time querying and continuous monitoring.

What you get

We return an easy to read scorecard that includes a rating tuned to the customer's particular use case, a probabilistic confidence level, one or more snippets of evidence supporting the rating, and even a link to the underlying source for manual verification. This data can be returned as machine readable JSON or can be rendered as a HTML scorecard in a customer’s own application or in the Trooly portal.

Use Cases

Trooly provides a breakthrough product for use cases in trust & safety management, risk management, and customer relationship management. Unlike legacy background, credit and risk checks, Trooly ratings require minimal customer data input and cost less than $1 per query, with ratings and supporting evidence returned in roughly 30 seconds. Trooly’s predictive power has been demonstrated to be at least twice as effective as background checks, credit checks or database profiles in determining trustworthiness of individuals and businesses. Primary applications for our product include:

Marketplaces and communities

  • Ensure that users and community members are trustworthy at the time of signup
  • For sensitive use cases, monitor users continuously for occurrence of negative events

Financial Services

  • Verify the provided identity and application traits including employment, education & income.
  • Screen customers for KYC/AML compliance purposes
  • Understand the risk and value of thin file or low FICO customers, including millenials, immigrants, and small businesses

Recruiters and Marketers

  • Understand the risk and value of applicants and customers for your particular organization's needs
  • Proactively identify potentially highest value, lowest risk recruits and customers for proactive sourcing

How it works

Trooly's Instant Trust ratings evaluate multiple dimensions of trust - mining real-time data to evaluate identity, personality and behavior against well-established academic and commercial models. These trust ratings are based on patented innovation* and breakthrough computer science research in areas such as information extraction, entity resolution and machine learning. Ratings can be configured to meet different trust requirements based on company, industry and product dynamics. In about 30 seconds, Trooly's platform performs the following six steps:

Input processing

We check the validity of input information, check that we have permission under FCRA to process the input, and perform numerous proprietary identity maps to develop a robust and compliant input identity.

Data gathering

In real time, we gather digital footprints from four types of data sources: the public web, the dark web, permissible social media, and our own aggregation of digital public records. At this stage, we typically get thousands of pieces of data per input identity that we have to process.

Identity matching

Our platform has outstanding identity matching models that allow us to match an input identity to a digital source with high confidence. We are able to do fuzzy matching with high probabilities, which gives us a huge advantage over legacy solutions which rely only on exact matches to huge amounts of intrusive input identity information.

Information extraction

In addition to ensuring a digital record matches an input identity, we make sure that we understand the context of a digital source so that we extract only what is relevant to the input identity. Once we are done with cutting edge identity matching and information extraction, thousands of pieces of data usually boil down to a handful of relevant ones.

Trait scoring

With the relevant extracted data, we run over 30 machine learned models to score the input identity on a series of trust-related traits learned from ground truth data from our customers, proprietary ground truth data that we have gathered independently, and from a range of academic and commercial models.

Holistic scoring

At this final stage, we configure, combine and weight our underlying traits in a way that best predicts the target behavior that our customer is seeking to understand.

Request a proof of concept

*U.S. patent awarded June 30, 2015 (US 9,070,088 B1): "Determining Trustworthiness and Compatibility of a Person."