|Dan Baker Blog|
Customer Analytics: Making the Strategic Leap From Hindsight to Foresight
“The right merchant is one who has the just average of faculties we call common sense; a man with a strong affinity for facts, who makes up his decision on what he has seen. The problem is to combine many and remote operations, with the accuracy and adherence to the facts so as to arrive at gigantic results without any compromise of safety."
—R. W. Emerson, Wealth from The Conduct of Life (1860)
In recent years, the analytics software market has become a colorful and vibrant Turkish bazaar. And it’s a delight to have a wide variety of analytic tools to choose from. Yet as we all know, the decision to buy software must be carefully weighed to ensure that the insertion costs from license, training, process change, and maintenance generates a good return on the investment.
And what I’ve just said only half states the problem. For in the real world, a carrier lives with scarce resources and must therefore learn to be a wise merchant. The money it spends on analytics software to reduce operational errors or retain customers is necessarily limited. There are many other revenue-generating priorities to be considered, such as rolling out new networks and services to the broader base of customers.
Well, getting advice on such decisions is the very reason why telecoms hire consulting firms. The consultant’s value is to hone in the core need – data intelligence to drive better decisions – and completely detach that from the decision to hire more people or buy more software, hardware and services. In fact, the best strategy may be to forego any purchase of software at all and redeploy existing assets and resources.
Frank Bernhard from Deloitte is a guru who enjoys pondering such grand strategies. In fact, in his role as director, strategy practice, in the telecom sector, Frank has consulted widely in analytics which is his specialty. So for those of you caught up in the tactical and firefighting issues of an average work day, it’s time to relax, kick up your shoes, and open your mind to some strategic thought.
Dan Baker: Frank, it would be great if you could explain how your team at Deloitte looks at service-provider analytics problems.
Frank Bernhard: Sure, Dan. First of all, within Deloitte, we have a very large dedicated strategy practice which I’m a part of. I lead the specific focus for how analytics and quantitative strategy apply to our clients inside the Technology, Media and Telecommunications practice or TMT. Our practice is geared around helping operators discover advanced uses of analytics.
We basically look at analytics as comprising three activities. The first part is hindsight, which is looking backwards and seeing what happened in the past and reporting. Hindsight is the low end of the analytics spectrum and in some cases it’s nothing but fancy reporting.
The second side of analytics is current insights or the here and now. An example here is real-time decisions. When someone calls to complain about poor wireless service, you need answers immediately about where the network is congested and whether or not this person is at risk to churn.
But where the real power comes in is the third side when you create predictive insights or foresight. This is a bit of the Holy Grail. It combines hindsight and current activity by the user in a way that when the customer calls, the operator can actually anticipate why they are calling.
Now while this sounds like sci-fi or very futuristic, the truth is this is all enabled by advanced statistics. Using techniques like logistic regression, we can start to understand and/or shape a user’s behavior. Pretty soon you can figure out what drives the person to churn or what elements of your calling plan or data plan offer are most attractive to the user.
DB: How about an example of using analytic foresight?
FB: An excellent example is a subscriber who recently acquired a handset from a mobile operator (Big Mobile) and a number of his calls were being dropped. He was frustrated enough with the poor quality service that he blasted Big Mobile with a harsh message on Facebook.
Now the Facebook complaint is one of many vectors that should warn Big Mobile that the subscriber is at risk. For instance, maybe the signaling records showed a high percentage of calls were dropped in the Metropolitan Statistical Area (MSA) where the subscriber lives and works. And maybe the subscriber visited a webpage that discusses early termination fees. The challenge for the operator is to bring those events and intelligence back in-house and understand that this is an early-warning signal of churn.
And the remedy is to be proactive and provocative at the same time. By “proactive" I mean doing something that would retain the customer on your network; and “provocative" — making sure that whatever additional offer you make exceeds the customer’s expectations.
Now an actual operator we work with is doing this advanced foresight analytics today. They are examining events from many data sources and within six hours a customer care person calls the subscriber to discuss the bad experience. The customer-care rep might tell the user that he noticed an unusual number of dropped calls have occurred in the customer’s area and that Big Mobile is expanding base stations in the MSA. And to compensate for the inconvenience, the operator might offer the customer a 10 percent bill reduction or boost the user’s allocation of minutes by 10 percent that month.
DB: What can operators do to move toward some of these advanced customer analytics programs?
FB: I think the starting point is figuring what kinds of data can help me become a stronger player in the market place? Or how are we going to woo our subscribers into the next level of spending?
One of the biggest problems operators face is they haven‘t structured their data in a way that you can perform meaningful analyses on it. There’s no shortage of data, but there’s no way to access it, codify it, and think about it in terms of a strategic outcome.
A few years ago I came up with something called the Analytics Continuity Model, a way of understanding the degree of automation or customization required by different analytics uses.
For example, the call center is at the low end of analytics needs. There’s a certain cringe factor when calling your telephone company. It’s usually not a pleasant experience because you have to explain why you’re calling, and the agents often don‘t have access to information they need to answer your concern. So this level needs a high degree of automation. An analysis of the customer’s profile and lifetime value should automatically trigger actions a call-center agent should take.
At the high end of the analytics continuity model, a greater interaction with the data is needed such as people working on marketing campaigns or folks trying to boost the spend of subscribers. Those folks ask complex analytics questions like: “If I stimulate a market with certain amount of television advertising spend, how much will that increase the overall life in the San Diego market?" Now to answers such questions requires quite a bit of analytics engineering and customization.
DB: Frank, thanks for this strategic thinking. As you know, many startup software companies have entered the analytics space in the past few years to capitalize on “big data" opportunities in telecom. What are your thoughts about this trend?
FB: Yes, there are lots of upstarts that have come into this market because they sense that there is money in it. So there’s a lot of excitement: “Oh great, let’s aggregate data and find a way to deliver synthesized reporting of it."
New entrants coming in are great as long as they fit into an operator’s overall strategy. It’s a matter of purposing that information and that doesn‘t always evolve from a small software strategy. If you talk to the telcos, even those mid tiers, the tier IIs and IIIs, they hesitate to use smaller startups because they fear the vendor will compromise their sensitive data. Then again, you can often still perform analytics on anonymous data, say, in the cloud, and then tie that back to real customer data that lives behind the operator’s firewall.
One trend I think will play out – across both software and consulting analytics – is for the solution and services to become far more specialized and granular. For example, I think that the channel-management strategy will have its own set of consultants and other solution vendors. Network operations and analytics is another area where a deeper set of vendors solutions and consulting expertise will emerge.
At Deloitte, we don‘t look towards the platform or analytics software solution to be the end all, we look towards the outcomes and the overall strategy to deliver the impact.
Everyone recognizes the hype and momentum around terms like “big data" and “in-memory computing," which is fine. But the question remains: "What are you going to do with all that unstructured data? How are you going to make it purposeful and deliver a measurable return in terms of people, technology, process, and business change?