In the latest Executive Insights video, supported by Mahindra Comviva, we examine how operators can unlock the goldmine of customer data. Here, MEF talks to Dirk Jungnickel, SVP of business analytics at du, about using data insights to reduce churn, create targeted offers and improve network performance.
For three decades operators enjoyed fantastic organic growth. At first, people simply wanted phones to make calls. Then they graduated to smartphones with data connections so they could browse the web, look at maps, listen to music.
Subscribers numbers boomed, ARPU grew.
But, as we all know, this natural boom could not last forever.
After a certain amount of time, there were simply no more people left to recruit. Then came competition. When the emphasis in telecoms moved away from cellular to IO/digital, that brought in new entrants. Thus, OTT players like Google and Facebook ate away at core services like messaging, while squeezing down margins.
Of course, this is business. Competition is inevitable. The challenge is to respond to it by analysing your strengths and building on them.
For many of the world’s operators, that means data and analytics.
Simply, telcos know a lot about their customers. Now, they are learning to use this data to provide better services back to customers, run the network more efficiently and even sell (in aggregated form) to third party verticals like retail and transport.
In the telco world, the customer continually uses our product. It’ a privileged position in that we can take that usage and analyse it. We have transparency over how they are using our product every minute – and what kind of product and service quality they are getting.
“It’s a strategic challenge that most operators in mature markets now face,” says Dirk Jungnickel, SVP of business analytics at du. “In the UAE we have a mobile penetration of 200 per cent. So we effectively can’t find any new human subscribers. And we’re under heavy pressure form OTT providers.
“So we are all looking for new revenue sources. We’re moving into completely new value chains in IoT, data monetisation, smart cities and so on. And analytics is a core element for all of these things.”
Jungnickel says the strategic advantage of mobile operators in analytics comes down to the breadth and accuracy of the data.
“We have an enormous amount of profile data on our customers: age, nationality, address. And we have customers who stay with us for years so we know a lot about their usage behaviour. But I think what makes us special is that we also have location data. The SIM card leaves a trail that tells us how a person moves through the country.
“I look at it this way: when you buy a book, you buy a book. But in the telco world, the customer continually uses our product. It’ a privileged position in that we can take that usage and analyse it. We have transparency over how they are using our product every minute – and what kind of product and service quality they are getting.”
As head of analytics at du, Jungnickel is investigating multiple avenues for exploiting this data vault. The most obvious is tailoring core services for the subscribers themselves. The two main aims? Getting people to stay with the network. Or to spend more.
Jungnickel says the data can say a lot about both. “We can build algorithms that the predict propensity to churn based on usage behaviour, spending patterns, nationality, age and so on,” he says.
“There are what we call champion variables that let you see these correlations. We use hundreds of these variables and then let the computers figure out the best predictor of churn propensity. Then you target them with propositions to retain them.
The really exciting component of all this is the ability of the algorithms to learn what works. This helps them to create highly targeted offers for small groups of customers. It’s like the systems are on auto-pilot.
“It’s the same for ARPU. You can build propensity models in a similar way for almost any product. You can calculate the probability they will accept the offer and then see what additional spend you’ll get.”
The really exciting component of all this is the ability of the algorithms to learn what works. This helps them to create highly targeted offers for small groups of customers. Jungnickel argues that it’s like the systems are on auto-pilot.
He says: “In a marketing function, you want to optimise maybe for several parameters – for revenue, retention and margin – or maybe a mix of all three. That makes this more of an optimisation problem than a prediction problem.
“The algorithms can try certain propositions for each customer profile in small samples, and continuously learn based on the responses. They can then optimise for your marketing targets: discounts on SMS or international voice or whatever. In principal it’s just a mathematical problem. The more sophisticated the predictive algorithms the better the results. It almost runs on auto pilot mode.”
Of course, another way to keep customers happy is to ensure the network is robust. Jungnickel says the same types of algorithms that predicts churn can be used to anticipate technical issues – just using different data sources.
This is fantastically valuable as it means du can move to a preventative maintenance approach. In other words, instead of sending out engineers on routine checks across the entire network, it can dispatch them to where failures are most likely to occur.
“We have thousands of pieces of equipment in the field, and maintenance is quite expensive,” he says. “If you can predict where maintenance is required before equipment breaks down, you can reduce cost and improve customer service quite a bit.”
There are equally big savings to be made in capex. Indeed, du is now using data analytics to deploy new equipment where it is needed most by looking at where, for instance, 4G users are most active. Jungnickel says: “Typically telcos in the Middle East spend 15 per cent of revenues on network capex. There are case studies that show you can reduce this to 10 per cent, which is a very significant saving.”