In today’s digital world, speed and adaptability often determine the difference between success and stagnation. In the final episode of this three-part series, Pierre-Yves Calloc’h, Global Chief Digital Officer at Pernod Ricard, sits down with Julian Schirmer, Co-Founder of OAO, to discuss what it really takes to deliver impact in complex, global data projects.

Navigating data challenges in global projects

In the third installment of the three-part series, Calloc’h delves right into the challenges around collecting data for optimization. He notes that while some consumer sales data existed, it was not organized in a way that supported decision-making.

In some markets, this lack of granularity made projects nearly impossible to sustain, says Calloc’h. “We couldn’t get all that data from the retailers in Australia. We’ve had to stop some of the projects for that reason.”

Historical data gaps: A roadblock for marketing optimization

One recurring obstacle has been the lack of sufficient historical data. For example, in Mexico, the team struggled because they did not have three years of data to work with. This gap forced them to pause certain initiatives, planning to revisit them in two years once adequate data was available.

There were also hurdles during the deployment of Salesforce in some countries. Normally, teams adopt about 80–85% of the system’s recommendations, but in this case, adherence dropped sharply to around 20%. The issue, as Calloc’h explains, was tied to geography. In markets with limited coverage, such as those spanning only a small area, the system worked well. However, in regions like Germany, where outlets are scattered across the country, the tool suggested impractical routes — in some cases, requiring up to 1,000 kilometers of travel in a single day.

To resolve this, the data science team refined the tool by adding a feature to group outlets geographically while still incorporating business rules. This adjustment, though complex, made the recommendations more practical and led to a successful redeployment of the tool.

Lessons from working in uncharted territory

According to Calloc’h, these challenges are part of the natural process of innovation. “That’s one of the things you need to do on those projects: you are going into uncharted territories, new places, and new ways of working.”

The key, he emphasizes, is to measure success consistently. “You need to be measuring your KPIs of success, the business impact and level of adoption, to ensure that you’re on the right track with the projects.”

Moving fast with MVPs: Managing uncertainty in projects

While doing an MVP, the first part is to find quick ways to the solution. Calloc’h notes that even if the data is not 99% accurate, one must get the data quickly, which requires reducing the parameter.

Instead of rolling out across a wide scope, he stresses the importance of narrowing the focus. “So it could be just one state in the U.S., but you go fast instead of solving everything for all the states in the U.S. Make sure that you get to the result and don’t go into the usual strategy in IT projects where you do the foundations first and then the first layer, etc. You cut through that.”

Because uncertainty creates resistance, Calloc’h says the priority should be demonstration rather than perfection. “You need first to prove that it works, and then you will invest in the foundations and the different parts, which are important for projects to run in the long term.” He further recommends doing whatever it takes to go fast.

Off-the-shelf or internal development?

Speed matters more than the perfect method and is a key criterion to go fast to actual usage. “So sometimes it might mean taking some off-the-shelf solution, even if you redevelop afterwards. Sometimes it’s faster to develop internally than to go through a process of request for proposal, then negotiation and contract signing, et cetera.”

Adding on, Calloc’h points to the startup mindset as a model. “So here, taking inspiration from the startups from the Agile methodology, et cetera, is key for those projects.” This approach is especially vital because in many cases, expertise is limited.

Ultimately, he argues that rapid prototyping is the best way to handle unknowns. “It’s unknown for a lot of people. And so the way to manage uncertainties is to go faster to the prototype and faster to the MVP, and this is when you prove the value, and then you can invest in the rest.”

The value of involving finance early

Reflecting on his experience, Calloc’h admits there is one thing he would change. “I would’ve involved the finance teams earlier. One of the things I’ve failed at is that by concentrating solely on sales and marketing, a valuable connection with finance was delayed,” he adds. “But we are creating new KPIs, which are interesting for finance. For example, in marketing, we are calculating our return on sales, and that’s super interesting for the finance team.”

The lack of early collaboration created extra work later, affirms Calloc’h, as he says, “They could have been involved more from the beginning with more adherence.”

Thinking beyond direct users

Moving forward, Calloc’h highlights that innovation does not stop with the primary users. He states that other users could also take advantage of some of the tools, the information, and the insights produced. This broader view helps expand impact and adoption. He then emphasizes the importance of partnerships within the organization and urges creating the right alliances from the start.  

Core skills for data and AI

Speaking of core skills, Calloc’h lists curiosity as number one, as the field is constantly evolving. He adds that success in this field goes beyond technical expertise and that the job is mostly about change management.

However, traditional training doesn’t always fit. Calloc’h adds that sometimes the organization finds different ways by engaging in workshops and developing games for people to understand in a simplified manner.

Big data, small data, and the need to explain the “why”

For Calloc’h, the work is a mix of big data and small data for people to understand the “why.” He shares a traffic jam analogy where knowing the why behind the jam makes one more patient. Likewise, the systems will give insights to people who want to understand the why.

Additionally, Calloc’h insists that fieldwork is indispensable to understanding what real customers are choosing and why. This practice grounds insights in reality and sharpens models.

Field validation also helps address errors, says Calloc’h. There might be an issue in data or modelling, and this helps to differentiate real insights from false positives. Concluding, he recommends being curious, taking change management into account, and ensuring a strong connection to real-life context.

CDO Magazine appreciates Pierre-Yves Calloc’h for sharing his insights with our global community.