Imagine any kid today trying to get a hold of a friend with the limited technology available when Gen Xers were growing up …
Back before there was such thing as a text message. Back when someone had to actually be home if you wanted to get them on the phone. Maybe someone knew someone that had a cell-phone, but it was a massive brick with no range or battery life to speak of, and the only pocket it would fit into was a pair of Jncos – but thankfully those hideous things hadn’t come on the scene yet either.
But things changed fast in the telecommunications industry. And with that change has come a flood of new data to manage, clean and interpret.
With a level of precision that almost feels like self-awareness, phones now know exactly where they are in the world– and telecoms save and track that data. Detailed call records reveal metadata that includes locations, durations, and destinations of calls – information that governments routinely track too. With more than half of all internet use now taking place on mobile devices, data usage results in an endless supply of information transmitted between phones and Internet servers that identify interests and consumer habits that marketing companies love to get their hands on.
All the capabilities have boosted demand, which in turn puts strain on the networks. Crowded spectrum and skyrocketing data traffic increase competition for telecommunications links.
And somewhere, buried in all that detailed information, are the answers to optimizing efficiency and delivering service as stable and reliable as POTS—Plain Old Telephone Service—landlines used to be.
The complexity and interconnectivity of these problems means job security for data scientists in the telecom industry who have come prepared for the challenge with a master’s level education.
Solving the Last Mile Problem with Data Science
Landlines haven’t gone away, of course. Without dedicated high-speed backhauls to serve cellular towers and other points of presence, the sexy high-tech wireless side of the business would be dead in the water.
Like utility companies, the greatest challenge for telecoms tends to be “the last mile”—the myriad, interdependent, idiosyncratic connections between trunk lines and the individual customer. And like utility companies, telecoms are switching to smart metering technology to monitor and improve all the various components of that infrastructure.
Intelligent Site Management is the telecom version of the Smart Grid for cellular service sites. Remote monitoring technology ensures that antennas are powered up, computer equipment cooled down, and that data traffic flows smoothly. The telco central office is notified immediately in the event of a problem; reaching further into the grid, devices may be redirected to other cells if the closest one is experiencing difficulties.
Creating the algorithms that allow all this real-time monitoring and switching to happen so smoothly that customers don’t even get a blip of interference is an ongoing challenge for data scientists in the telecom industry.
A company called Napatech has approached the problem by integrating data from customer appliances and telecom network probes to allow real-time analysis of network issues. Providers implementing their solution have been able to use it to reduce packet loss and operate networks more efficiently on the fly.
Awash in Customer Data, Telecoms Struggle to Manage it Successfully
As we have all become more connected in our daily lives, telecommunications use has surged. According to a 2013 article in Antennas Online, global data traffic in 2012 exceeded that of all prior years combined.
And with all those texts, emails, search queries – and the occasional call – going out over the network, the amount of information that telecoms gather about their customers has exploded.
Once upon a time, when all calls were manually connected and snooping operators sometimes neglected to take themselves out of the circuit, local phone companies knew quite a lot about who was using their services and how.
When automated equipment began to take over in the early 1900s, that sort of intimate information was lost to telcos. They knew where service was installed, when and where calls were made, but little more.
Orwellian Dystopia … Maybe Someday, but for Now Just Marketing
Now, the system has come full circle. Big Data allows telecommunications providers to track every significant detail about every use of a communications device. In addition to the basic demographics collected when a customer signs up for service, telecoms now have access to detailed information derived from SMS, voice, and data usage patterns. In many cases, they can even harvest information from the device when it is not in use.
In the abstract, it seems obvious that you can tell a lot about a person from the people they interact with and the places they visit. But in practice, it’s a big job for a telecom data scientist to figure out how to analyze that data to tell them something interesting about you.
Although this degree of monitoring makes some people uncomfortable, and some see only nefarious conspiracies at work, the fact of the matter is that telcos mostly want this kind of data for one purpose: marketing!
Location, Location, Location
Telecommunications providers are not the only businesses that want to sell you something. But they are the businesses that know some of your most personal details, including the one coveted piece of information that every retailer would like to know: Where are you!?
Location-based advertising is a grail of sorts that telecom companies are hoping to offer to partners. As a 2014 article in AdWeek illustrates, advertising can be far more effective when presented in the right context. And that context is often predicated on location.
In New York, ice cream chain Van Leeuwen rolled out a location-based marketing application that displays coupons, deals, and directions to their stores when a customer passes by. Five percent of the chain’s revenues are now processed through the app.
Data Flows in Both Directions: Data-Based Telecom Marketing Solutions
Competition is fierce among telecommunications providers. The more they know about prospective customers, the better they are able to deliver the right services at the right price.
According to a 2012 McKinsey white paper (PDF), a surprisingly detailed profile can be developed from this customer data. These days, what they know about users can literally include what they had for lunch! Did you search for takeout on your phone? That search request went into a database somewhere!
That information is used in a Marketing Mix Model to help match products to customers. The data generated by the model reveals increasingly finely segmented slices of the customer base. With the proliferation of service offerings from telecoms, it’s vital to match the right offer to the right customer segment.
Customization: The Key to Customer Retention
Churn, or customer turnover between providers, is a major obstacle for the industry. Telecoms are doing everything they can with the data they have to try to retain customers.
Personalization is the key. The extent of these offers today might just be a bundle of cable, cellular, and landline services, but as the algorithms improve, data accumulates, and billing and provisioning systems become more advanced, it might result in a hyper-personalized service customized exactly to each individual customer’s usage patterns.
The ability to put customer preferences together and compare them to cost-of-service models could allow telcos to make you an offer on the fly that would beat current prices but still allow them to make a tidy profit on the deal.
Extracting and parsing this data to make the numbers work is a job that only a skilled data scientist is capable of performing.
Privacy Concerns and Legal Obligations
With all this detailed data floating around about their customers, telecommunications companies have to be sensitive to privacy concerns. Indeed, they are legally obligated to do so. The federal Telecommunications Act of 1996 mandates that telcos protect proprietary customer information.
Data scientists help devise methods to ensure that information remains secure, but they are also responsible for aggregating the data so that it remains useful.
Using a method called Next Product To Buy (NPTB), customer segment data in the abstract is run through Bayesian rules and market basket analysis to compute conditional probabilities about the products a member of that market segment is most likely to purchase next.
Such methods preserve individual privacy but still allow telecommunications providers to provide vendors with enough information to offer you products and services you might actually want or need. Even if only begrudgingly, most mobile device users take advantage of this kind of customized marketing in one way or another and would agree that the telecom industry has struck the right balance.