Debrief: 100+ Execs Contribute to Big Data Analytics Discussion
Author: Telecom Council of Silicon Valley
On July 23, 2014 over 100 telecom executives convened for a ½ day deep dive on analytics (meeting photos here and attendee list here). It’s no secret that trends towards cloud computing, parallel processing, virtualization, flash memory, and cheap storage have aligned and delivered a new ability to pour through millions of data points in real time, and to pull out actionable business knowledge. This meeting allowed the professionals most deeply involved with the advancement of analytics to delve into what’s already been done, and what opportunities and threats remain for this booming segment.
*Of special note to Telecom Council members, full access to all presentation materials are available via the member-only library.
Expert speakers invited to present to the group included Mark Evans, a Research Scientist for Sprint’s Advanced Analytics Lab, Ashish Gupta, VP of Marketing at Actian, Ben Parker who is the Principal Technologist at Guavus, Marc Bensadoun who is CEO at Newfield Wireless (a Tektronix Communications Company), Sameer Merchant the Vice President of Commerce Services, Audience Network & Payment at Opera Mediaworks, Scott Sobers the Director of Marketing and Strategy Communications Service Providers at Teradata, and Alexander Russakovsky who is VP of Technology in the Digital Telco Architecture Group at Huawei. As is typical with all Deep Dives at the Telecom Council, we also included 9 pre-filtered startups that are creating particularly interesting solutions in this segment. This meeting’s Innovation Rapid Fires were presented by DropThought, Jolata, Skyfire, Neo Technology, Argyle Data, Zementis, StreetLight Data, SQLstream, and Parstream.
With such a big topic at hand, we started our discussion at the collection point of much of this data –with a telco. Mark Evans, Research Scientist from Sprint’s Advanced Analytics Lab, highlighted their journey through the new “gold” – Big Data. He reviewed the process from optimization and research, to a Hadoop-based environment, and enhancing the technical capabilities of the enterprise data warehouse. Big Data for telcos, historically, was utilized for business optimization – a top down approach. Sprint began an internal research challenge, to turn this approach upside down. The bottom up approach would let data do the talking, creating new opportunities.
The initial proof of concept for this challenge was a transition from optimization to asking questions about the data. By using some old servers and the open Hadoop platform, Sprint was able to learn:
- New Analytics: Using network data to create new products, increase customer satisfaction, and attract new customers by providing actionable insights to customers and enterprise decision makers
- New Benefits. New Data: The power of incorporating new, insightful data sets, open source tools, new specialized business rules, and geospatial techniques, as well as an examination of new business intelligence, and visualization tools.
- New Infrastructure. Lab Cluster: Retreaval, storage and analysis of data was going to require changes to the infrastructure, so Sprint conducted trials of distributions & server setups (selected open source solution), training of internal resources (Big Data User Group), and expanded the teams to run prototypes on the cluster (usage based cost / finance, application data transforms / product, location accuracy improvement / network, pathing analysis / marketing, device behavior analysis / device, customer text analytics / care).
- New Infrastructure. Production Cluster : the next step was integrating standard visualizations and analytics tools, and proving funding use cases. They also expanded on IT process & controls related to continuous data loading, transformations, and reliability, standards established, resources scaled – they migrated from a team of 5 supporting the lab cluster to more than 5 teams responsible for the system.
Mark’s introduction to a true telco case study set the stage nicely for the meeting’s keynote by Ashish Gupta from Actian, a software company specializing in Big Data solutions. Ashish reviewed the progressions of Big Data -from Big Data 1.0, which was primarily based on CRM data and was supposed to address churn (but mostly gave good deals to customers who were going to retain anyway) to Big Data 2.0. We migrated from single-sourced data to multiple, integrated sources which provide rich, easy to use, industrialized, real-time and affordable options which drive broader use. Thanks to the progression of Big Data analytics, we’ve advanced from a 5% churn prediction using CRM data to being able to predict around 80% of churn, by adding use of other devices, CDR and location data.
Next to the stage were our panelists from Guavus, Newfield Wireless, Opera, Teradata, and Huawei, who participated in an interactive discussion with the audience about Big Data telco customers. Huawei’s Alexander Russakovsky started the discussion with a list of 5 common Big Data mistakes:
- Over-abstraction: building independent layers and ignoring efficiency
- Over-virtualization: virtualizing everything at the expense of performance
- Data Dumping: storing data before figuring out how it will be used
- Functionality Fragmentation: rubber bands integration of unrelated products
- Ignoring data mobility/life-cycle characteristics and dependencies within data
Those highlighted mistakes weren’t protested since participants, panelists, and telcos alike agreed on one point throughout the meeting – complexity. As illustrated via the image above, the network complexity promotes different points to use this valuable data. True not just for the complexity of the network and data, but also for the actions created by that data. Every bit of analytics used mushrooms into more data and more opportunities which creates more actionable intelligence.
While the benefits and applications are easier for the customer and provider to see, there are also many concerns surrounding the Big Data debate – security and identity among them. Current security solutions sit on the perimeter. It was a consensus that the right kind of solution will sit at the origin of the data. Apple and Google manage relationships with 1.5+ trillion users. Are the telcos using that data to avoid the dumb pipe? Where does that data get secured?
Another concern which continued to come up throughout our discussion was that of multiple users. The intelligence offered currently doesn’t know if mom is using her device or if it’s her toddler. How can we make our intelligence more intelligent and bring identity into the picture more often? Although there was clearly plenty of opportunity remaining in this area, panelists contributed use cases where analytics were used to track clusters of users more than device specific tracking. Also noted, was all the added sensors being added to our devices which help the provider understand anomaly detection, which is the most common way to understand if your device is under control of the primary user or not.
By lunch time, participants reviewed some incredible progress in this area, but much of the discussion surrounded all the outstanding opportunities in the space… Calling all communications innovators; now is the time.
We finished up the day’s discussion by welcoming Meredith Attwell Baker, the new CTIA President and CEO, to the stage. Joined by Liz Kerton, President of the Telecom Council, and Jon Metzler, President of Blue Field Strategies, Meredith participated in an open discussion with the audience on the topics of spectrum and broadband. Baker’s predecessor, NFL Hall of Famer and former Congressman Steve Largent, had a penchant of opening with football anecdotes. Baker acknowledged she had big cleats to fill, and very clearly and quickly articulated how she plans to do that, through representing the mobile ecosystem more broadly on policy issues, not just wireless carriers. In essence, she described a much bigger tent for CTIA. Given that the layers in mobile have decoupled in the smartphone / post-smartphone era, and given that service providers will have their own regulatory issues to confront, this makes total sense.
What was most striking is how she described this in about 60 seconds. At a time when mobile is no longer an industry but an underlying capability, and when, mobile means not just carriers, but equipment vendors, service providers, developers, enterprise, and yes, consumers, this resonated. And at a time when multiple government agencies are asserting jurisdiction over mobile, this may be a timely development.
Interested in joining the next Council discussion? See what’s next here – http://telecomcouncil.com/eventlist.php
Have something to add to these notes? Contact Palmer (at) telecomcouncil.com.