Philip Morris International [NYSE: PM]

David Rodrigues

Data Value Steward

David Rodrigues is the Data Governance & Data Management for Brazil at Philip Morris International. He has over 20 years of experience working with data at Fortune-500 companies.

Rodrigues shared his expert insights for the 2025 edition of CIOReview on generating value with data and AI, while producing the desired business results.

An apparent contradiction exists today between the immense investment in AI and the low concrete results for most corporations. On one hand, every week new technological advancements are published, while on the other, C-suite executives are frustrated with the lack of real impact on their businesses. During my over 20 years of experience working with data at Fortune-500 companies, I identified three elements that, when combined, generate exponential business results: firstly, a deep connection between the data team and business strategy; secondly, the development of data products that work in practice, and thirdly, the capacity of the data team to communicate effectively with stakeholders.

Elements Generating Exponential Business Results

First and foremost, the chief element to generate value with data and AI is a deep and continuous connection with business strategy and consumers. It is paramount that the data team includes a role in general known as the translator. These people not only have technical skills, but also a deep understanding of the business processes. They establish a trusted and personal relationship with key business partners, such as the CEO, COO, and CCO, and participate in all high-level strategic discussions, bringing a strong point of view and challenging the status quo. They also spend time with consumers to understand their pains and expectations. The translators guarantee that the data team identifies the most important business problems to solve, and articulate the right data and AI solutions to address them.

 Generating value with data and AI is less about being the best technologist and more about the way the data team navigates the organization and articulates how data and AI solutions resolve real business problems. 

Together with the connection to the business, the data team must provide data solutions that work. In this context, "to work” means to be fit for purpose, to consistently produce results that make sense. Let me illustrate that with a failure, a real example at a company I worked for before I joined Philip Morris International. Back then, one of our main business challenges was to build a model to anticipate the impact of price changes on market share. We built a very sophisticated model, with state-of-the-art technology, leveraging the best solution available in the market. However, after several tests, stakeholders kept questioning the results, which did not reflect reality. Every time an inconsistency appeared, the team tried to calibrate the model, but errors kept surfacing. In the end, stakeholders lost trust in the solution, and we had to abandon the project. While investigating “post-mortem”, we concluded that the problem was weak data quality and governance, but it was too late. There were problems with how data was collected and transformed, and it was not fully clear who owned the data we were using. Therefore, when I say that a data product must work, it is not about having 99% accuracy, but in producing consistent, reasonable, and useful results that build trust among stakeholders and enable decision-making.

The third element, which works in combination with the other two, is the communication that the data team establishes with the organization. That includes two aspects: being able to explain in simple terms the logic behind how the data solutions work, and the capacity to showcase the tangible business outcomes that these data products can produce. For instance, years ago, after developing a consumer segmentation model, my team organized a workshop with real consumer participants, pre-screened with the segmentation algorithm. During the event, our C-level executives had the opportunity to interview and talk directly with real consumers to understand their pains and expectations. The workshop brought the results to life, and for years, executives remembered the real people they had talked to. The event created vivid memories in participants' minds and helped build strong trust in the model and its results. Thus, developing good models is not enough. The data team must also have the capacity to explain in plain English how data solutions work, and to show how they impact business results in practice.

To conclude, generating value with data and AI is less about being the best technologist and more about the way the data team navigates the organization and articulates how data and AI solutions resolve real business problems. This comprises three elements: first, the capacity of the data team to connect with business strategy, consumers, and key stakeholders; second, to develop data solutions that consistently generate results that make sense, and third, the capability of communicating the logic behind the models in simple terms and of showcasing their impact on the business. With that said, in order to deliver the value the business needs, the data team must get out of their own offices and spend time with consumers and business stakeholders. Being present in important discussions and feeling consumers' and stakeholders' pains and expectations is the best way to gain perspective. As they say, a problem well stated is a problem half-solved.