To understand the EthicsGrade research process, consider three tiers of input into our model and the quality assurance of our data.

At the core, our vision is that ESG data can and should be automated in its production, and so the Machine Learning engine we are building is being designed to as to make our own research process as efficient and therefore scalable as it can be. There is no magic to this, and all our methods we’d be very happy to discuss with clients. Our goal is simply to make data collection a machine-driven task, and ensure that our human expertise is applied to questions of judgment, review and expanding the scope of our research.

The second tier of our process is our analyst team. During the phase of our development where we conducted ratings manually, we developed a ‘Red Team’ concept in order to ensure that the analyst could defend their assessment of a company and any subjectivity was surfaced, and rigorously debated. As we move towards fully automated ratings, our analyst team’s role is shifting to becoming the ‘Red Team’ for the machine. Analysts will continue to train our Machine Learning model on new areas that we bring into scope of our model, but the main focus of their work will be to review the analysis that has been automated and to correct its mistakes.

The third tier of our process is our advisory panel. Led by co-founder Dick Nodell, this group is charged with two tasks. First, to be the point of escalation for any challenge that is brought to EthicsGrade as to the quality or validity of our data. There can be only three logical heads to such a challenge, that we considered the wrong question, we evaluated the wrong evidence, or we weighted the data point wrong. Our panel is there to consider these questions and iterate our model where appropriate. We also believe in ‘eating our own dogfood’. A key requirement that we look for in organisations that we rate is that they move from ‘principles’ to ‘pronouncements’ i.e. actionable and fully articulated decisions derived in a consistent manner through the use of ‘protocols’. Any challenges that we have brought to our Advisory Panel will be reasoned in a transparent manner and the outcomes will be published in the public domain. This will give the companies we rate confidence that they have an opportunity to challenge where we have erred, and customers of our data confidence that we are consistent in our approach.

The second function of our Advisory Panel is to expand the scope of our model. We currently look at a narrow set of questions that we deem to be important for questions of ‘AI Governance’ and ‘AI Ethics’ in particular. This is unsatisfactory to us, as we know there is a bigger debate around questions of Corporate Digital Responsibility that we want to address, as well as big social issues that urgently need attention such as Diversity and Inclusion that are impacted by technology and therefore a natural extension to our work. Our Advisory Panel will curate a set of Working Groups to look at the model expansion. Please contact us if you would like to register your interest in joining one of these expert groups.

The Advisory Panel is operationally independent of EthicsGrade, and EthicsGrade commits to be bound by its decisions.

For more information on the qualitative aspects of our model, please visit