It’s no secret that communication service providers (CSPs) are under a lot of pressure as they make massive investments in upgraded networks while facing shrinking margins and revenues from their eroding traditional voice or broadcasting businesses.
Traditional operators understand that they must go beyond what they did before. They need to offer more compelling services to reduce churn and acquire new customers. But how to know what services customers want most, and how much to charge for them?
A key asset CSPs have is the huge amount of information that they generate and maintain. And so it's the analytics from their massive data sets that becomes the go-to knowledge resource as CSPs re-invent themselves.
Our next Big Data innovation discussion therefore explores how the telecommunication service-provider industry is gaining new business analytic value and strategic return through the better use and refinement of their Big Data assets.
To learn more about how analytics has become a business imperative for service providers, peruse this interview with Oded Ringer, Worldwide Solution Enablement Lead for HP Communication and Media Solutions. The discussion is moderated by me, Dana Gardner, Principal Analyst at Interarbor Solutions.
Here are some excerpts:
Gardner: What are the major trends leading CSPs to view themselves as being more data-driven organizations?
Ringer: CSPs are under a lot of pressure. On one hand, this industry has never been more central. Everybody is connected, spending so much more time online than ever before, and carrying with them small devices through which they connect to the network. So CSPs are central to our work and personal lives—as a result, they’re under lot of pressure.
They’re under a lot of pressure because they’re required to make massive investments in the networks, but they also need to deal with shrinking margins and revenues to subsidize these investments. So, at the end of the day, they’re squeezed between these two motions.
One approach many CSPs have adopted in the last year was to reduce cost and to cut operations. But this is pretty much a trip to nowhere. Going into most basic services and commodity services is no way for these kinds of things to survive.
In the last two to three years, more and more traditional operators understand that they must go beyond what they did before. They need to offer more compelling services to reduce churn and acquire new customers. They need to leverage their position as a central place between consumers and what they are looking for to become some kind of brokers of information.
The key asset they have in their hand to become such brokers is the huge amount of information that they maintain. It’s exactly where analytics comes into play.
Gardner: When we say CSP and telecommunication companies these days, we’re more and more talking about mobile, right? How big a shift has mobile been in terms of the need to analyze use patterns and get to know what's really happening out in the mobile network?
Ringer: Mobile services are certainly the leading tool in most operator’s arsenals. Operators that have the subscriber 'connected' with them wherever they go, around the clock, have an advantage over those that are more dependent upon or only provide tethered services.
But we need to keep in mind that there’s also a whole space for analytics solutions that are related to fixed-line services like cable, satellite, broadband, and other, landline services. CSPs are investing a lot in becoming more predictive, finding out what the subscriber really wants, what the quality of those services are at any given time, and how we can reduce churn in their customer base.
Another kind of analytics practice that operators take is trying to be predictive in their investments in the network, understanding which network segments are used by more high-worth individuals, those that they do want to improve service to, beefing up those networks and not the other networks.
Again, it’s these mobile operators who are on the front lines of doing more with subscriber data and information in general, but it is also true for cable operators and pay-TV operators, and landline CSPs.
Gardner: Oded, what are some of the data challenges specific to CSPs?
Ringer: In the CSP industry, Big Data is bigger than in any other industry. Bigger, first of all, in volume. There is no other industry that runs this amount of data—if you take into consideration they’re carrying everybody’s data, consumer and enterprise. But that’s one aspect and is not even the most complicated one.
The more complicated thing is the fact that CSPs, unlike most enterprises, need to handle not only the structured data that’s coming from databases and so on, but also unstructured data, such as web communication, voice communication, and video content. They want to analyze all those things, and this requires analyzing unstructured data.
So that’s a significant change in that type of process flow. They are also facing the need to look at new sets of structured data, data from IT management and security log files, from sensors and end-point mobile device telematics, cable set-top boxes, etc.
And two, in the CSP industry, because everything is coming from the wire, there’s no such thing as off-line analytics or batch analytics. Everything needs to be real-time analytics. Of course, this doesn’t mean that there will not be off-line or batch analytics, but even these are becoming more complex and span many more data sets across multiple enterprise silos.
If you analyze subscriber behavior right now and you want to make an offer to improve the experience that he’s having in real time, you need to capture the degradation of service right now and correlate it with what you know about the subscriber right now. So it's so much more real time than in any other industry.
We’re not talking here about projects of data consolidation. It may be necessary in some cases, but that’s not really the practice that we’re talking about here. We’re talking about federating, referring to external information, analyzing in the context of the logic that we want to apply, and making real-time decisions.
In short, CSP Big Data analytics is Big Data analytics on steroids.
Gardner: What does a long-term solution look like, rather than cherry picking against some of these analytics requirements? Is there a more strategic overview approach that would pay off longer term and put these organizations in a better position as they know more and more requirements will be coming their way?
Ringer: Actually we see two kinds of behaviors. The market is still young. So it's very hard to say which one will be more dominant. We see some CSPs that are coming to us with a very clear idea on what business process they want to implement and how they believe a data-driven approach can be applied to it.
They have a clear model, a clear return on investment (ROI) and they want to go for it and implement it. Of course, they need the technology, the processes, and the business projects, but their focus is pretty much on a single use case or a variety of use cases that are interrelated. That’s one trend.
There’s another trend in which operators say they need to start looking at their data as an asset, as an area that they want to centralize. They want to control it in a productive manner, both for security, for privacy, and for the ability to leverage it to different purposes.
Those will typically come with a roadmap of different implementations that they would like to do via this Big Data facility that they have in mind and want to implement. But what’s more important for them is not the quickest time to launch specific processes, but to start treating the data as a central asset and to start building a business plan around it.
I guess both trends will continue for quite some while, but we see them both in the market—sometimes even in the same company in different organizations.
Gardner: How does a CSP really change their identity from being a pipe, a conduit, to being more of a rich services provider on top of communications?
And what is it that HP is bringing to the table? What is it about HP HAVEn, in particular, that is well suited to where the telecommunications industry is going and what the requirements are?
Ringer: HP has made huge investments in the space of Big Data in general and analytics in particular, both in-house developments, multiple products, as well as acquisitions of external assets.
HAVEn is now the complete platform that includes multiple best-in-class product elements based on multiple, cutting edge, yet proven, technologies, for exploiting Big Data and analytics. Our solution for the space is pretty much based on HAVEn and expanded with specific solutions for CSP needs, with a wide gallery of connectors for external data sources that exist within the CSP space.
In short, we’re taking HAVEn and using it for the CSP industry with lots of knowledge about what traditional CSP operators need to become next-generation CSPs. Why?
Because we have a very large group, within HP, of telecom experts who interact with and leverage what we’re doing in other industries and with many of the new age service providers like the Amazons, Googles, Facebook and Twitters of the world. We go a long way back in expertise in telecom—but combine this with forward thinking customers and our internal visionaries in HP Labs and across our business units.
Gardner: Just to be clear for our audience, HAVEn translates to Hadoop, Autonomy, Vertica, and Enterprise Security, along with a whole suite of horizontally and vertically integrated set of applications that are vertical industry specific. Is that right?
Gardner: Tell me what you do in terms of how you reach out to communications organizations. Is there something about meeting them at the hardware level and then alerting them to what these other Big Data capabilities are? Is this a cross-discipline type of approach? How do you actually integrate HP services and then take that and engage with these CSPs?
Ringer: Those things exist, like engaging at a hardware level, but those are the less common go-to-market motions that we see. The more popular ones are more top-down, in the sense that we are meeting with business stakeholders who want to know how to leverage Big Data and analytics to improve their business.
They don’t care about the data other than how it’s going to result in actionable intelligence. So, at the CSP level, it can be with marketing officers within the CSP who are looking to create more personalized services or more sticky services to increase the attention of their subscribers. They’re looking to analytics for that.
It can be with business-development managers within the CSP organization that are looking to create models of collaboration with the Yahoos and Facebooks of the world, with retailers, or with any kind of other participants of their ecosystem where they can bring the ability to provide the pipe, back-end hosting of services and intelligence about how the pipe is providing the services and the sentiment of the customers on the other end of the pipe.
They want to share information of value to their customers, making them dependent on them in new ways that aren’t just about the pipe thereby gaining new revenue streams. That’s the kind of motivation they have. It can be with IT folks as well, but at the end of the day the discussion about CSP Big Data isn’t coming from the technology. It’s coming from the business people that understand that they need to do something with the data and monetize it.
Then, of course, it becomes pretty quickly a technical discussion that the motion is business to technology, rather than infrastructure to technology.
We also developed the support practice within our organization that does exactly that—business advisory workshops. It’s for stakeholders of different roles to realize what the priorities are in using Big Data. What is the roadmap that they want to implement?
The purpose of this exercise is to quickly bring everybody to the same room, sit together for a day or two, and come out with an agreement on how to turn themselves from conventional services to more personalized services and diversify the business channels via using information data.
For several years now, we have one large customer, Telefónica a Latin American conglomerate, has been working with us on analytics projects to improve the quality of experience of their subscribers.
In Latin America, most people are interested in football, and many of them want to watch it on their mobile device. The challenge is that they all want to watch it during the same 90 minutes. That’s a challenge for any mobile operator, and that’s exactly where we started a critical project with Telefónica.
We’re helping them analyze the quality of experience. Realizing the quality of the experience isn’t a very complicated thing. There are probes in the network to do that. We can pretty accurately get the quality of experience for every single video streaming session. It’s no big deal.
Analytics kicks in when you want to correlate this aggregation of quality with who the subscriber is, how the subscriber is expected to behave, and what he’s interested in. We know that the quality isn’t good enough for many subscribers during the football game, but we need to differentiate and know to which one of them we want to make an offer to upgrade his package. What’s the right offer? When’s the right time to make the offer? How many different offers do we test to zero in on the best set of offers?
We want to know which one of them we don’t want to promote anything to, but just want to make him happy. We want to give him a better quality experience for free, because he is a good customer and we don’t want to lose him. And we want to know which customer we want to come back to later, apologize, and offer him a better deal.
Based on real-time triggering of events from the network, degradation of quality with information that is ongoing about the subscriber, who the subscriber is, what marketing segment he belongs to, what package is he subscribed to and so on, we do the analytics in real time, and decide what the right action is and what the right move is, in order for us to give the best experience for the individual subscriber.
It’s working very nicely for them. I like this example, first of all, because it’s real, but also because it shows the variety of processes we have here with correlation of real-time information with ongoing information for the subscribers. We have contextual action that is taken to monetize and to improve quality and to improve satisfaction.
This example touches so many needs of an operator and is all done in a pretty straightforward manner. The implementation is rather simple. It’s all based on running the right processes and putting the right business process in place. But this isn’t always straightforward for enterprise customers, particularly those in the small to medium enterprise segment, so imagine what CSPs could do for their customers once they’ve gotten a handle on this for their own businesses.
Gardner: It seems to me that that helps reduce the risk of a provider or their customers coming out with new services. If they know that they can adjust rapidly and can make good on services, perhaps this gives them more runway to take off with new services, knowing that they can adjust and be more agile. It seems like it really fundamentally changes how well they can do their business.
Ringer: Absolutely. It also reduces quite a lot the risk of investment. If you launch a new service and you find out that you need to beef up your entire network, that is a major hit for your investment strategy. At the same time, if you realize that you can be very granular and very selective in your investment, you can do it much more easily and justify subsequent investments more clearly.
Gardner: Are there any other examples of how this is manifesting itself in the market—the use of Big Data in the telecommunication’s industry?
Ringer: Let me give another example in North America. This is an implementation that we did for a large mobile operator in North America, in collaboration with a chain of retail malls.
What we did there is combine their ongoing information that the mobile operator has about its subscribers—he knows what the subscriber is interested in, what their prior buying pattern and transactions were and so on—with the location information of where the individual person is at the mall.
The mall operator runs a private wi-fi network there, so he has his own system of being able to track where the individual is exactly within the mall. He knows within two meters where a person is in the mall but with the map overlay of the physical mall and all product and service offerings to the same grid.
When we know a person is in the mall, we can correlate it with what the CSP knows about this person already. He knows that the specific person has high probability of looking for a specific running shoe. The mobile operator knows it because he tracks the web behavior of the specific individual. He tracks the profile of the specific individual and he can have pretty good accuracy in telling that this guy, for the right offer, will say yes for running shoes.
So combining these two things, the ongoing analytics of the preferences, together with real-time location information, give us the ability to push out targeted and timely promotions and coupons.
Imagine that you go in the mall and suddenly you pass next to the shoe store. Here, your device pops up a message and that says right now, Nike shoes are 50 percent off for the next 15 minutes. You know that you’re looking for Nike shoes. So the chance that you’ll go into the store is very good, and the results are very good because you create a “buy-now or you’ll miss-out” feeling in the prospect. Many subscribers take the coupons that are pushed to them in this way.
Of course, it’s all based on opt-in and, of course, it’s very granular in the sense that there are analytics that we do on subscriber information that is opted in at the level of what they allow us to look at. For instance, a specific person may allow us to look at his behavior on retail sites, but not on financial sites.
Gardner: Again, this shows a fundamental shift that the communications provider is not just a conduit for information, but can also offer value-added services to both the seller and the buyer—radically changing their position in their markets.
If I am an organization in the CSP industry and I listen to you and I have some interest in pursuing better Big Data analytics, how do I get started? Where can I go for more information? What is it that you’ve put together that allows me to work on this rather quickly?
Ringer: As I mentioned before, we typically recommend engaging in a two-day workshop with our business consultants. We have a large team of Big Data advisory consultants, and that’s exactly what they do. They understand the priorities and work together with the telecom organizations to come up with some kind of a roadmap—what they want to do, what they can do, what they are going to do first, and what they are going to do later.
That’s our preferred way of approaching this discipline. Overall, there are so many kinds of use cases, and we need to decide where to start. So that’s how we start. To engage, the best place is to go to our website. We have lots of information there. The URL is hp.com/go/telcoBigData, that’s one word, and from there you just click Contact Us, and we’ll get back to you. We’ll take you from there. There are no commitments, but chances are very good.
Gardner: Before we sign off, I just wanted to look into the future. As you pointed out, more and more entertainment and media services are being delivered through communication providers. The mobile aspect of our lives continues to grow rapidly. And, of course, now that cloud computing has become more prominent, we can expect that more data will be available across cloud infrastructures, which can be daunting, but also very powerful. Where do you see the future challenges, and what are some of the opportunities?
Ringer: We can summarize four main trends that we’re seeing increasing and accelerating. One is that CSPs are becoming more active in enabling new business models with partnerships, collaborations, internet players, and so on. This is a major trend.
The second trend that we see increasing quite intensively is operators becoming like marketing organizations, promoting services for their own or for others.
The third one is more related to the operation of the CSP itself. They need to be more aware of where they invest, what’s their risk and probability of seeing a specific ROI and when will that occur. In short, Big Data and Analytics will make them smarter and more proactive in making the investments. That’s another driver that increases their interest in using the data.
Overall they all look to become more proactive, they all realize that data is an asset and is something that you need to keep handy, keep private, and keep secured, but be able to use it for variety of use cases and processes to be ready for the next move.