"Act like a real digital company..." Why Vodafone is all-in on the data cloud
Vice President, Global Telecom Industry, Google Cloud, Google Cloud
Enterprises around the globe have been working to gain more insights from their data and create a more intelligent, data-driven approach to uncovering value. Like the Hair Club before it, Vodafone, the global communications service provider, is not only building the networks that provide others with the data and connectivity to transform—it’s become a leader in digitally transforming itself so that others may follow suit.
Alberto Marco, Global Head of Analytics & Data Services, and Dr. Cengiz Ucbenli, Global Head of the Big Data and AI Group, are two of Vodafone’s leaders behind this work. They recently joined George Nazi, VP for Telco, Media & Entertainment solutions at Google Cloud to discuss how to build big data that’s not monolithic; achieving faster execution while reducing costs; and how “there is no local IT and group IT—we are now all IT.”
George: Vodafone is one of the largest operators across the globe. You serve upwards of 300 million users in 24 countries on three continents. Carrying three terabytes of data every day. This is no small task, managing across so many scenarios and jurisdictions. Where do you even begin?
Alberto: Your question is about solving data sprawl and the scale of data. Historically, we’ve been very much localized in terms of our data. Now, we have a clear blueprint for transitioning that to a centralized concept. We’ve moved to a data ocean, as we call it, with a very clear blueprint on how to segregate data, per market and per service.
Now the risk is, we could have built this massive, monolithic data lake. What we’re trying to do instead is build data domains around specific types of data and specific data products. So you have a network domain, a commercial data domain, a financial domain, and an operational data domain. And all of those sit within that taxonomy that we built.
What we want to do is enable self-service and agility, but with the underlying layer of privacy and security built in.
Cengiz: When operating in so many countries, there’s many factors at play. Standardization becomes key, and not just standardized data. We need standardized platforms, a standardized tool stack for our teams, standardized privacy controls and regulatory controls, all working across all of these markets.
Thanks to Google Cloud, we are making great strides in standardization so that we can act like a real digital company instead of building separate models for each market. We can build once, centrally, and deploy many times and make it scalable. That’s one of the biggest benefits from shifting to the cloud.
George: With this work underway, how is Vodafone looking to leverage data to improve customer experience?
Cengiz: We wanted greater granularity on what was affecting NPS, our net promoter score, with customers. Maybe they were on price plan that didn’t suit them, or looking for better indoor network performance—we couldn’t really be sure. What we are doing now is more data-driven, more comprehensive, looking at all the anonymized and aggregated data points to understand, down to the individual level, what customers really value.
Alberto: We’re also looking at smart planning to optimize our infrastructure, as well as looking at things like network anomaly detection, to catch problems before they can happen or grow. There’s a lot of areas enabling us to improve network coverage such as root-cause and gap analysis and more advanced coverage maps, gap analysis. Another area is customer journey, and using predictions to ensure they have the best possible end-to-end experience.
George: With so many potential inputs, attacking the data problem is complex. Organizationally, where does Vodafone start, and how do you prioritize?
Alberto: As recently as five years ago, Vodafone had about 17 petabytes of data on 600 physical servers, and 1,300 pipelines. By the end of some strategic programs we’re working on, we will have 11,000 data pipelines going onto one central cloud-based platform powered together with Google Cloud.
On one level, it’s practical. Our original platform had reached its physical capacity to grow, with more than 700 servers. The need to maintain and renew and update that—it would take well over 12 months to even get new hardware in the data center, let alone to get it to work, and as soon as you did, investments from three or four years ago became redundant.
On another level, it’s strategic. With cloud, there’s agility and flexibility, but also energy saving for sustainability and cost. Cloud is about the constant release of new technology. When you get new tools that come in, if you just have a rigid infrastructure, it’s very hard to bring in new technologies. With cloud, it makes it very easy to be first adopters of new technology and therefore gives us competitive advantage.
George: Where else are you seeking competitive advantage?
Cengiz: We’re hiring data scientists in a way we never did before. We have dozens under the central commercial organization who handle projects across the company, and then dozens more in Italy, Germany, the UK, Turkey, South Africa, Egypt—they’re everywhere. We call them the Big Data AI Team.
George: What’s an example of what they’re doing?
Cengiz: We have three big pillars the team is working on, and cloud will play a big role to achieve that strategy. One pillar is real-time AI. We need to understand our customers’ pain points. We need to make decisions in a timely manner when we are extending our network, introducing new services and offering new price plans. Instead of making predictions on a monthly basis, we are now able to do it on a real-time basis.
An additional pillar is AI at scale. I believe that Google Cloud helps standardize the infrastructure, so this can open the door to scalability. And the third pillar is AI everywhere. We want to find a use case for AI in every specific business function. It could be the network, it could be operations, it could be pricing. Everywhere.
This is where some of the Google Cloud tools like Vertex AI come in very handy. For example, there is a big investment by Google or other technology companies in speech recognition or video recognition. We don’t want to take on these additional investments on our own. We can plug-and-play for our own unique commercial use cases.
Vodafone has three main pillars for its data-driven network: Real-time AI, AI at scale, and AI everywhere.
George: You’re a few years into this journey. What are some core business benefits you’re seeing and lessons learned?
Cengiz: First, faster execution and information flow across multiple markets. This is what I expect from our cooperation with Google Cloud. We talked about scalability of the use cases. We are standardizing all of these machine learning pipelines so we can reap the benefits of Vodafone’s global scale—build once, deploy many times—and reducing overall time to market across our many regions all at once.
The second is cost reduction. As Alberto mentioned, we save money by not having to maintain largely inflexible big-data platforms. But we got rid of that when we moved over to Google Cloud. I would also expect to see some accuracy improvements in our models, which has also translated into new revenue generation.
The last one are those prefabricated models, more off-the-shelf tools. Instead of us making that investment, maybe we can bring in plug-and-play solutions.
Alberto: If you go on a journey to the cloud and you start moving data there, you need to build a foundation of success. It’s about having the right people and the right skills around cloud, data, security, and data literacy. Then you build that data “factory.” Your people can then manufacture products and the analytics on top.
To be clear, we’ve been doing this for over five years. Vodafone has been linking analytics to marketing and commercial outcomes in multiple countries for quite a while. But what we’re now doing, it’s not about the killer app. It’s that we can manufacture multiple killer analytics apps at industrial scale. And our data factories are key to us in how we can build those data products and analytical products to give us a competitive advantage and allow us to pivot. Hence, we adopted cloud so that we can build common platforms that everyone can use.