The world’s communication service providers (CSPs) are sitting on more big data than perhaps any other industry, with hundreds of millions of devices and massive networks. Many are looking to apply artificial intelligence (AI) to help them exploit the massive amounts of data they collect and to better monetize their networks. In addition, they are seeking to create new and improved customer experiences, thereby radically transforming the economics of the CSP business. But the big question remains for a lot of CSP CIOs: How do we get there from here?
To help answer that question, here’s what we’ve observed and learned from our CSP customers who’ve successfully applied AI and analytics to big data to meet their business goals. Their AI and analytics journey and recommendations could be very valuable for your business as well.
- Decide What You Want To Accomplish And Which Data Will Get You There
It seems obvious, but it’s important to put this stake in the ground in terms of what you want to accomplish. Some business problems and objectives are bigger than others and thus it’s critical to decide which top business problems or objectives you want to address and what success looks like.
For example, one Asian mobile network operator (MNO) we work with, with extensive fiber and mobile networks throughout its home country, has aimed high. First, they had their sights set on entering all-new markets. Their long-term objective was to become a one-stop-shop supplier to subscribers in its country along the lines of the Amazon business model. Shorter term, though, they were looking for insight into real-time quality-of-experience (QoE) levels for each customer and to learn what over-the-top (OTT) services they were using.
From a data analytics standpoint, the MNO wanted real-time visibility into subscriber experiences so they could take steps to ensure that those experiences were always top notch and thus attract and retain customers. Second, they wanted to create highly targeted new services and promotions based around other services their subscribers were accessing using their network so they could add new revenue streams. And finally, they wanted to sell demographic data to online advertisers based on their subscribers’ behavior. By establishing their goals and identifying the necessary data to help them reach them, the operator is now providing a better customer experience and new services that are helping to expand their business.
- Consolidate All Your Data In One Place
At the outset, it can be hard to predict where and how subscribers are using the network. Part of the problem for CSPs can be the rise of OTT apps, such as video services, which customers subscribe to separately but use their CSP’s network to access.
Petabytes of data flow through CSP networks every day, and you may want to use that data to monitor and better understand the QoE levels. By doing so, MNOs can view usage data so you can turn it into customer-impacting decisions. But a big challenge can be getting all the data into one place to improve data access performance. It’s important to make the investment in a large data lake so that data can be accessible immediately — not pulled, sorted, then analyzed slowly from different systems. All data should flow into a high-performance data lake so actions and decisions can be made in real-time.
But then what? A challenge is getting a pipeline to that data lake to run analytics on it.
- Integrate Your Data Lake With Your AI and Analytics
For most CSPs, it’s been difficult to integrate their data storage repositories with modeling and analytics platforms. The traditional process has involved building interfaces from each of the various repositories to each data science platform, one at a time – this is costly, complex and time consuming . The other option for a CSP, which is what my company offers, is to automate the analytics process, delivering a highly instrumented experience driven by use cases that are relevant to the MNO business. By integrating the multiple analytics use cases with a centralized, scalable data lake architecture, this will accelerate results, enable real-time decision making and cause market disruption.
- Learn how to “fish” for your future
We often use the “teach a man to fish and he’ll eat forever” approach. In other words, we teach our customers everything we know about AI — the ‘ins and outs’ of not only our solutions but what we’ve learned working with other CSPs deploying AI/ML analytics over the past decade, so that customers can eventually be self-sufficient and harness their data in many different new ways far into the future.
If you’re a CSP just getting started, the first phase of your analytics project may be to create a micro app — for mobile network data, for example. Below are the steps you might take:
- Develop key use cases and desired outcomes
- Create a multi-step process to emulate operator behavior
- Conduct workshops on best practices and knowledge sharing across stakeholders
- Normalize the data
One CSP we recently worked with had more than 20 data sources just for voice over LTE (VoLTE) data alone. Getting all this data normalized and verified was important to their achieving their goals. Many people think you simply can throw data into an analytics system and it just works. But data normalization is key to the integrity of the results. We conducted a number of fact-finding and discovery workshops with our customer as part of the normalization process.
As a result of their AI and analytics journey, the Asian operator I mentioned earlier is now able to enter new markets, generate new revenue with advertising sales, and upsell and recapture market share.
Today, they participate in the digital retail banking market and have entered public safety, supplying local governments with connectivity to streetlights with panic buttons and security cameras. They are now able to match customer-identifying information and demographics to their subscribers’ mobile usage behavior, based both on geography and the URLs they visit. As well, they’ve able to combine that information in ways that allow them to sell advertising for new revenue streams. And they can now see QoE levels for every network subscriber. This data, in part, has helped them go to work to capture share in the booming video services market.
Key Best Practices
As I mentioned earlier, CSPs have more data and more complex networks than any other industry and they have a lot they can teach the rest of the world about how to use AI for success. In my experience, the CSP CIOs seeing the most success with AI and analytics:
- They are visionary in their approach and open-minded to possibilities of the seemingly impossible.
- They are are not afraid to make mistakes.
- They partner with companies with similar vision and with proven experience to handle very large CSP networks.
- They have a robust IT infrastructure with the ability and willingness to quickly deploy needed resources, and a dedicated team to quickly remove roadblocks to progress.
Having the agility to get things done and keep things moving is critical to winning with big data analytics and AI.