Speed and relevance are critical for a modern corporation’s survival. But if you’re as big as a major telecommunications firm, how do you achieve them?
You can definitely be quick when you’re small, because you’re probably not facing the same complexities. You can also be fast when you’re big. But if you steam off in the wrong direction, you’re in big trouble.
Major corporations often float in place because of this. They cannot agree internally on what to do next “analysis paralysis” while nimble new firms are working to sink them.
These large firms have a new tool; however, that actually leverages their scale for speed. It’s Big Data.
Big Data can eliminate problems before they happen and that’s the fastest speed you can have.
I speak to Fortune 100 firms every day. They are eager to learn what we know about Big Data analytics. They want to know how they can unlock their data, to increase the speed and intelligence of their critical decisions.
It’s a new paradigm. The massiveness of a corporation represented by mounds of data and tentacles in every direction – is a liability that can be turned into an asset by Big Data.
If you are reading this magazine, you know that Big Data is the ability to find predictive insights from processing data in four V’s: large volume, great variety, lightning velocity and proven veracity.
Suddenly, those servers full of business information are not baggage, they are fuel.
Take the service-assurance measurements across our network things like remote sensors. We’re digesting 30 billion data points every hour.
What if we could use this type of data in new ways to keep our business fast and fresh, and adopt insights rapidly from headquarters down to the front line?
Here are some examples:
• We use streaming analytics tool to dynamically monitor the load on our network and balance the traffic between adjacent cell sites in real time. In the old days, we adjusted cell site parameters once a month.
• In our stadium distributed antenna systems, small cell sites are now adjusting parameters every 15 seconds. We can shift capacity literally when everyone moves out of their seats to the bathrooms at halftime.
• A tool called the Tower Outage and Network Analyzer, developed by AT&T Labs, helps us make big gains in outage recovery times. It uses real-time data, combined with historical patterns of previous disturbances, to predict the ripple-effects of various repair options. People never know or care that data’s getting crunched in a massive way. They just know their phone is working fine.
"Big Data can eliminate problems before they happen and that’s the fastest speed you can have"
This is not just about making smartphone users happy a particular concern in telecom. Having a better, faster network also enables the Internet of Things. Our major customers are able to gather Big Data from all points of the globe to make better, faster decisions.
Here is a Big Data/Internet of Things example that could apply to anyone.
Like many major corporations, we have a significant fleet 75,000 vehicles. Things break, and traditionally we’ve then focused on fixing them to get our technicians back on the road.
Last year we took a new approach, collecting up to 1,500 sensor measurement per vehicle every three or four seconds. Starting with batteries, we wanted to figure out how to predict when they would fail before they did.
Our model was up and running in four months, and we’re showing around 10,000 fewer vehicle breakdowns needing tows or jump-starts per year. More of our customers get on-time service at their homes, which drives continued business on top of the annual savings.
On battery failure prediction alone, we’ve applied models that reach 80 percent accuracy. And now that we have built a platform, we are able to quickly produce other predictive maintenance models for other parts of the vehicle.
Here’s another example that could apply to any major company: customer care. Typically, the largest customer-facing corporations have the hardest time making sure the care experience is good, because we have millions of interactions. But Big Data can help.
We are analyzing online chats discovering what makes customers happy or frustrated. This is a massive unstructured data set, and there is no way to analyze it without Big Data tools. Now we can identify success patterns and create more speed to resolution. We can help determine if the issue is a particular training course, a particular supervisor, etc. We can conduct data-based A/B testing for the best customer outcomes.
So major companies like us are seeing the light. It’s really a “big three”: Make better decisions, drive out costs, and serve customers better.
Have there been roadblocks? Absolutely.
We evaluated more than 30 companies in a search for enterprise scalable software in support of data management and data sourcing. The lack of such applications keeps data inaccessible to those who need it to generate value.
We found very few creditable, scalable products. And this is basic plumbing for Big Data discovery of data, movement of large data sets and data cleansing.
So we’ve built some of it ourselves. For instance, we’ve created a capability to help us find and inventory data automatically across our many systems, applications and structures. Now we continually re-inventory across the company in a Hadoop environment.
We’ve also put several of our capabilities into open source for anyone to use, working closely with companies like Hortonworks. These include RCloud – a social code-sharing environment and Nanocubes which delivers Big Data visualizations through a web browser on a laptop. We’ve also introduced a high-quality streaming analytics tool called Tigon jointly with a firm called Cask.
We believe that Big Data is an outstanding way to make an asset out of something that could be a liability. Big Data “accelerates acceleration” because each project speeds the next.
It offers predictive maintenance, network optimization and a host of speed-based opportunities to telecom companies that will make our customers happier with us … without even knowing it was the data that did it.