PEP data quality: ComplyAdvantage’s 12-eye review process explained

Accurate and reliable information is the bedrock of effective politically exposed person (PEP) risk management, as misidentifications or omissions can have severe legal and reputational consequences. 

In this article, Madalina Morar, International Affairs Research Analyst at ComplyAdvantage, explains the company’s internal review process regarding its PEP data, highlighting the work that goes on behind the scenes before a PEP profile appears on the user’s platform. 

Data quality checks

From identifying a PEP to the point where the entity is searchable on the ComplyAdvantage platform, two main processes occur:

1. The international affairs research analysts (also known as the IARA team) research and select institutions and PEP-relevant positions.
2. The data strategists team collect the data and ensure it is regularly updated. 

Within these two processes, a series of inbuilt quality checks are implemented to ensure that the data is comprehensive, robust, and provides the best risk insights.

Each review starts with an audit of existing data for the researched jurisdiction. This includes the evaluation of existing data and setting the parameters of the research based on the company’s internal taxonomy. 

PEP categorization

Following this, the researcher completes a background search on the governmental structure of the jurisdiction and maps it according to the ComplyAdvantage taxonomy. The taxonomy is structured in four main pep-classes based on the FATF definition: 

Once the researcher has validated the relevance of the institution, where it belongs in the ComplyAdvantage taxonomy, and which public positions within that institution are PEP-relevant, a different researcher reviews the findings. 

At the end of the review process, one of the team members reviews the research for the jurisdiction in its entirety to make sure the coverage is comprehensive.

Data collection

During the data collection stage, the data strategy team utilizes page scraping tools to create individual scrapers that gather data from webpages identified by the research team. While one data strategist prepares the scraping agent, two other analysts review the data quality to ensure the accuracy of extracted positions and the comprehensive collection of all relevant data points, such as a photo, date of birth, and start date (if available). 

Monitoring information sources

Regarding source websites, the vast majority are primary sources, i.e., the information is taken from the website of the researched institution. If not available, other official sources such as governmental websites or aggregate governmental transparency platforms are also reliable sources. An important consideration in choosing a website for scraping is the possibility of scraping it again in the future in a consistent manner so the latest data is always utilized. In a limited number of cases, especially for countries with lower levels of transparency, the IARA team also retrieves PEP information from press releases and appointment decrees.

The 12-eye review process

Once the collection process has been completed, scrapers are scheduled to run automatically at preset frequencies. Data strategists and researchers will only intervene when:

• The structure of the website changes.
• PEPs can no longer be found on the page
• Human input is otherwise required. 

In each case, every new ingested source is reviewed by no less than six people from the research and data strategy team, reducing the risk of human error. If the automated scraping tool is reviewed due to changes on the website, it will be handled by an additional one or two people resulting in further quality checks.

On top of regular automated updates on sources, the researchers team monitors the national level parliamentary and presidential elections, appointments of cabinets, cabinet reshuffles, coup d’etats, and other changes in the composition of national level top legislative and executive bodies. Once the new PEPs are published, the IARA team updates ComplyAdvantage’s database within the following 24 hours.