More and more organisations are transforming their business models to become data and AI-ready. Data and AI-readiness mean organisations’ goal is to utilise the collected data to tap into the algorithmic monetisation of business activities and operations. This transformation is a big decision and concerns the entire organisation and its way of working.
To justify such a change, organisations need to have a clear value proposition that is sustainable over time. At the same time, organisations must adopt practices to get that value from data and AI activities. That is not an easy task for most organisations. According to Gartner’s recent research, only 20% of organisations claimed explicitly that they would allocate that value.
The key to sustainable value management of any Data and AI transformation, according to Jo Coutuer, Chief Data Officer at BNP Paribas Fortis, are practices around Discovery, Ensuring Efficiency, Setting Priorities, and Valorisation.
In the interview with him, we spoke about how organisations derive value from data and AI and overcome any limitations they may face. What trends should organisations know about data management and value realisation? What are the benefits for organisations of value measurement and monitoring? Let’s explore.
Hyperight: Can you please tell us more about you? What are your professional background and current working focus?
Jo Coutuer: My professional background is mostly one of entrepreneurship in the field of Data and AI. After a few initial experiences in Telco, Government and Big4 Consulting, I co-founded a data and business intelligence company in 1999. In 2012, the business was acquired by Deloitte Consulting and I continued to serve my very loyal clients as a partner at Deloitte until 2016. 2016 happened to be the time when globally systemically important banks (GSIBs) came under pressure to manage their risk and finance data in a more systematic way, in the context of the Basel regulation.
That is when I joined BNP Paribas Fortis, the biggest Belgian bank, as its first Chief Data Officer and Member of the Executive Committee. That role consists of three key activities, being the more defensive aspects of Data Governance and Privacy, the more operational aspects of Business Intelligence and Big Data and the innovative activities of Analytics and AI. Having completed a 6-year tenure as a Chief Data Officer, I have decided to go back to my entrepreneurial roots in the very near future.
Hyperight: During this year’s Data 2030 Summit, you will share more on “Why rigorous value management is key to sustainable data and AI transformations”. What can the delegates at the event expect from your presentation?
Jo Coutuer: I will present to my audience a tale of Three times Three. In the first part, I will give three arguments why they should care about knowing the value of their data and AI activities. In the second, I will share three practices I believe I got right during my tenure as a Chief Data Officer. And most important, to conclude, I will share three practices that I believe I should have spent more time on. I’m sure the last three may prove to be the most valuable as I am convinced that many of my colleagues are at risk of neglecting them as well.
Hyperight: There is an increasing need for organisations to create value out of data and AI during their data and AI-driven transformation journey. What does value management in data and AI projects mean, and why is this important for organisations?
Jo Coutuer: According to a Statista survey in 2020, only 40% of companies consider that they monetize data through their products or services and only 20% claim that they explicitly account for that value.
To manage value, four core processes need to be pursued rigorously and repeatedly. These activities revolve around Discovery (finding the right things to do), Ensuring Efficiency (doing things with the right quantity of resources), Setting Priority (doing things in the right order) and Valorisation (agreeing, measuring, accounting and communicating about value).
These processes are important because they are already embedded in the use of other, more traditional production factors, such as people, money, equipment etc. If data professionals dream of ever achieving a solid and sustainable level of productivity and recognition, then they urgently need to adopt similar practices. I am not just saying this to accommodate for the question that Davenport raised in 2021 in the Harvard Business Review, being “Why do Chief Data Officers Have Such Short Tenures”.
I am saying this to address the root problem. The root problem is that we – data professionals, whether we hold corporate roles or consulting roles or whether we are providing technology solutions, often have one thing in common: We seem to be obsessed by the “what” and the “how”. The “what” is often called “the use case”. The “how” is the technology, the architecture, the dashboard, the algorithm, the data lake, the data mesh, etc.
By being overly focused on the “what” and the “how”, we – data professionals – risk speaking in a different format than that of our fellow CxO’s who tend to be more focused on the stakeholder objectives.
In order to achieve impactful results, alignment with the more mature disciplines is crucial. A key element in that journey to maturity is to start from the stakeholder objectives, understand what defines value and then implement the aspects that I mentioned before as part of the Data Strategy Execution.
Hyperight: Your presentation emphasises that value management for data and AI is essential to success. But at the same time, you acknowledge that value articulation in data and AI projects is not simple and is challenging to reach. What are the ways available for organisations to derive value from data and AI and overcome any limitations they may face?
Jo Coutuer: First of all, nothing in business is ever simple. Let me take an example from public infrastructure development. It is not simple to express the value of building a new bridge or a new harbor dock. Nevertheless, it is an accepted and essential practice to articulate and quantify the direct as well as the indirect positive and negative effects, both on the economy as well as on various aspects of the environment. If doing that is possible – albeit not an exact science – then what would be the excuse of the data professionals to not identify the stakeholder objectives, how value is expressed, what direct or indirect value should be generated and how it will be measured, monitored, accounted and communicated?
The point that I am making here is that the first prerequisite to generating value is to identify what “value” is and in which “currency” it is expressed and to then do the intellectual effort of measuring it. The problem is rarely that the data and AI solutions fail technically. They are more often perceived as failing because they have either pursued the wrong or unknown value drivers or have failed to measure and communicate their impact. For the interested reader who is looking for a fairly generic and not too complex categorization of business value drivers, I can also refer to the article “Business Strategies for Data Monetization: Deriving Insights from Practice” by Julius Baecker et al., from 2020.
Hyperight: The monitoring of the value in data and AI projects is required. What are the benefits for organisations of value measurement and monitoring, and how can they perform value monitoring?
Jo Coutuer: The benefits are clear. Allow me to name just a few. Actively seeking out the stakeholders and their top level objectives keeps you from getting lost in nice-to-have use cases and will give direction to your multi-year data and AI strategy. Once you have a strategy, you will need to prioritize your investments in terms of financial as well as intellectual capital. Prioritizing must happen along the dimensions of expected value, regardless of how value is expressed (financially or other). When dealing with longer projects, the value question needs to be re-addressed, as drivers of value may have changed. A project may need to pivot and focus on new value objectives. Hardly any data or AI project should be delivered without real-life value testing, for instance through A/B testing. Nobody likes to abandon a fully delivered data-baby, but we tend to forget what the run cost is of data and AI efforts. If a data product does not pass the value test at the time of initial delivery, organizations need to have the guts to abandon or pivot. During the life cycle of a process, which uses AI or data products, value should be periodically reaffirmed. Value can leak away either because of technical reasons, such as model drift, or environmental factors that lead to data-product irrelevance.
Hyperight: In an organisation, it’s the Chief Data Officer’s responsibility to identify and create value of data. What are your recommendations and advice for the CDOs when we discuss value management in data and AI projects?
Jo Coutuer: Thank you for this provocative question. I both agree AND disagree with your statement. I could argue that the CDO is not alone in creating value. That would be like saying that the manager of a gold mine in South-Africa is responsible for the profitability of Mercedes in Germany, just because gold is an indispensable material in the equally indispensable computer chips that run a modern day car. On the other hand, it IS the CDO’s responsibility to make sure he or she is mining and providing the appropriate materials and that he or she understands, aligns and is part of one single value chain, all the way up to the car dealership, or in this case, the data products in the hands of decision makers or integrated in other software components.
I believe I have repeated my recommendations multiple times throughout the previous answers, so I will not repeat them.
Hyperight: During data management value creation, many decisions need to be made by organisations and the leadership. What steps should organisations be careful about to succeed and not fail or learn fast from the failures?
Jo Coutuer: Do we believe that data management is so different from the management of other production resources? I plead for an alignment of data management and governance practices with other management and governance practices in the company. There is absolutely nothing to be gained from putting “data” and “AI” outside the regular practices and processes. So in order to succeed, focus the energy on being rigorous and maturing the data practices, instead of believing that data and AI innovations alone will save the world.
Hyperight: How do you see the future of data management and the data management and value realisation of organisations? Do you see any trends in the upcoming years?
Jo Coutuer: It is my prediction that a lot of organizations will go through Gartner’s infamous “trough of disillusionment” when it comes to data and AI. Especially the ones that have overhyped data and AI without having put in place value monitoring and rigorous value based portfolio management. A second trend is that investors and financial markets will start to inquire actively about the expected returns from AI. The value of an AI portfolio will become a factor in or even a proxy for the public and private valuation of companies. So company stakeholders should stay ahead of both the disillusionment as well as the investors by professionalizing their own data and AI value management. After all, nobody can express and appreciate the stakeholder value better than the company’s own.
Content on similar topics:
- Activate Your Data. Accelerate Business Value – From Raw Data to Active Intelligence – Adam Mayer, Qlik
- Accelerating “data to value” with data scientists, to meet future transportation need – Robert Valton, Volvo Group Connected Solutions