Methodology

Sanctify ESG in short

Number of companies

13.000+

Countries/regions

USA, UK, Nordics and the major European countries

Languages

English, German, French, Swedish and Dutch

History length

Since 2010

Categories

E, S, G, ESG and 27 sub categories based on a well established framework

Number of news sources

55.000+

Number of news articles

25.000.000+

Number of news articles processed per day

~20.000

Number of data points

8.000.000.000+

Benefits

Fully automated

An automated system that runs around the clock without stop that can be scaled to fit the need.

Updated on a daily frequency

Our system keeps up with the latest news and is always up-to-date.

AI-based models

AI can be trained to perform complex tasks that otherwise would require human involvement.

Unbiased

  • There is no external influence on our scores and the same methodology is applied to all companies to ensure fair assessment and to make scores comparable between companies and over time.

  • No preconceptions about companies that clouds the judgement when analyzing them.

Independent

Scores are based on information from third party sources and not on company’s own communication which makes it more trustworthy.

Transparent

  • A user can look at a single article and see how it’s scored individually. Our score time series are derived from the scores of the individual articles.

  • Transparent architecture built by several smaller modules where every module has a specific purpose that is easy to understand, making the whole methodology less of a black-box.

  • We provide accessible documentation of our system.

Scoring Overview

Collect news from over 55000 sources. We download and store around 20000 new articles every day. Our database goes back to 2010.

Our system has multi-language support

We currently support the following languages English, German, French, Swedish and Dutch.

Assess all news articles linked to a company and verify that it actually is related to the company. Only articles that have been verified as related to a company will be used for further processing.

Run each news article through our sentiment analysis and categorization models. Each article gets a sentiment which is represented by a numeric value. Each article is also put into a category, or none if it’s not considered to be related to any.

There are subcategories to E, S and G which are not illustrated in this image.

Aggregate the sentiment score for all news articles into a score time series. This is done separately per category.

Download the full scoring documentation

Download the full scoring documentation to learn more, or contact us to book a demo.