Lost in translation? The position you need to fill to ensure that your AI projects are successful

According to the research and consulting firm Gartner*, some 50% of companies will struggle to bring their artificial intelligence (AI) projects to fruition. The most frequently cited reasons are a lack of AI skills among staff, insufficient knowledge of the benefits and limits of AI, and challenges concerning data (e.g. data silos). 

AI is a team sport

Many companies have already set up analytics teams and departments that bring together a range of different skill sets. The success of AI projects often hinges not just on the capabilities of data scientists, but also on agile, interdisciplinary teams that cover the broad spectrum of required profiles, including visualisation experts, data engineers, machine learning engineers, user experience designers and representatives from all areas of the business. One absolutely vital role in such teams that is commonly overlooked is that of the analytics translator.

Analytics translators: linking data science and business

The term “analytics translator” was coined by McKinsey in a 2018 article in the Harvard Business Review (HBR). Analytics translators play a decisive role in translating challenges faced by the business into AI use cases. A key part of the job description is being at home in both banking and data science, speaking their respective languages and knowing what makes them tick. Analytics translators must have a good understanding of the business as well as a working knowledge of statistics and data science and the ability to interpret and apply the concepts involved.

What does an analytics translator do?

A translator helps the company to ask the right questions at the start of an AI project and to identify problems that can be solved using AI technologies. A common mistake is starting out by focusing on a particular new technology and then trying to find applications for it. This is rarely a successful approach. It’s much better to start with the business’s needs and challenges and then find the technology that fits them.

Another important task for analytics translators is to prioritise use cases accurately and direct the available resources towards those that will have the biggest impact. Impact shouldn’t be measured in purely financial terms, it should also include factors such as customer needs and acceptance, innovation, probability of successful realisation, regulatory requirements and scalability.

Another central success factor is having analytics translators provide end-to-end support for AI projects that continues at the very least until the application transitions to the productive environment. Even with productive AI applications, it’s important to measure success on an ongoing basis and to continually develop the application wherever possible.

What skills and specialist know-how do analytics translators need?

Sector-specific expertise is one of the core requirements for the job. In the financial sector, this includes detailed knowledge of the regulations. As far as data and AI projects are concerned, the most important are bank-client confidentiality and data protection law. Analytics translators must also have excellent technical knowledge and a good command of the latest AI, data science and statistical methods.

Experience of various project management methodologies (conventional, agile and hybrid approaches) is also helpful in AI projects. Since AI projects often involve high levels of innovation, it can also be very useful for analytics translators to have an entrepreneurial spirit and experience of, for example, design thinking and value proposition design.

The SBA’s guidelines on handling data in day-to-day business can help analytics translators

At the start of June, the Swiss Bankers Association published its guidelines on handling data in day-to-day business. They cover many of the challenges mentioned above and offer aspiring analytics translators a good overview of the legal, technical and organisational requirements relating to data and AI projects in the financial sector.

Banks aren’t just asking what kinds of analysis and AI applications are possible with the data at their disposal, they also have ethical questions and challenges to consider with regard to using data responsibly. The guidelines also address these burning issues with some useful recommendations.

*the research and consulting firm Gartner

About the author

Christian Diethelm-Spiss is Head of AI Corporate Banking at UBS Switzerland AG. He co-authored the Swiss Bankers Association’s guidelines on handling data in day-to-day business.

Blog series on handling dataDigital Finance & Cybersecurity


Christian Diethelm-Spiss
Head AI Corporate Banking, UBS Switzerland AG