Connected Planet is working with STL Partners to provide coverage of the first of the U.K.-based strategy consultant’s influential Telco 2.0 events to reach the U.S., to be held Dec. 9-10 in Orlando. As a lead-up to that event, STL CEO and founder Simon Torrance and his team shared with us — and we’re sharing with you — some analysis of the most recent Telco 2.0 event, held last month in London.
Up today: Insights into Telco Data 2.0, a session that explored the issues surrounding telco resources of subscriber data, the opportunities they create, the barriers to success and the critical requirements.
Session presenters included Greg Skibiski, CEO, Sense Networks; Phil Laidler of Telco 2.0; and Michael Sandoval, CEO, Atigeo. Tim Sefton, Customer Director, O2 UK; Martin Hayward, Director of Strategy, Dunnhumby; and Paul Magelli, Head, Subscriber Data Management, Nokia Siemens Networks, joined the panel.
1. Telco data is much more interesting than it sounds
And the ways it’s interesting aren’t always obvious. Sense Network’s Greg Skibiski described how it was possible to model postpaid customers, where operators usually have much more data, and use methodologies like principal-components analysis to determine which signals in the data that exist on all the customers correlate with useful metrics, to validate the model against known data - and then extract similar metrics from the more limited prepaid OSS data as well.
2. Mining the CDR pile is technically non-trivial
This requires a different package of technologies and skills than telcos usually have. SQL databases don’t scale well for datasets as big as these, especially not when this is coupled with computationally complex processing. Instead, the Google approach is appropriate – operators should be looking at technologies like Hadoop/MapReduce, the so-called NoSQL databases, and cloud computing.
3. But the rewards are very real
Surprisingly basic data from telco operations can reveal rich social information. If you want to identify the “young and edgy” demographic among your customers, you can identify them because they go out and leave their home cell at night. Business travelers are likely to spend money and need value-added services when they go somewhere they haven’t been before.
And operators always want to know who’s going to churn. It turns out that the main predictor of churn is living near and calling other churners – churn is transmitted by word of mouth, and these metrics are a proxy for their social networks. A crucial point is that it’s much easier to start with internal applications like churn reduction and move on from there than to start by trying to make this a value-added service.
4. But it’s not your data, it’s your customers’ data
Phil Laidler of Telco 2.0 laid stress on the risk of a privacy disaster like the one that would occur at T-Mobile (which “lost” data for its Sidekick users) just after the conference. He argued that the role of the telco was as a custodian of subscriber data, whose value had been created by the subscribers themselves. Telcos needed to be aware that they do, indeed, have a lot of customer data – specifically, they have many more categories of customer data concentrated in their databases than other businesses.
5. Aggregation and the 80/20 rule
As a result, it’s vital for operators to get good at anonymising, aggregating, and pre-processing the raw data, in order to make it a deliverable service and to render it non-toxic before letting it out of the building. This is probably a good thing in commercial terms, as the bulk of the benefit from targeted marketing comes from identifying demographic groups – the problem with the “segment of one” is that diminishing returns set in as you attempt to target ever smaller groups, and the risk of losing user tolerance goes up rapidly as the targeting gets more individual.
6. Gmail vs Phorm
The upshot is that much depends on the launch strategy. Telco 2.0’s Phil Laidler used Google Mail and Phorm as examples of success and failure; where Phorm and its customers initially tested the system on subscribers without their knowledge, Google used invitation-only marketing to recruit power users. Google could also claim to be offering direct benefits to the end user, and its “permanent beta” strategy meant that the public could work out their own relationship with Google Mail over time and that Google could apply a continuous improvement methodology to work out the bugs.
7. Bad Customer Data
Atigeo CEO Michael Sandoval concentrated on identifying the causes of failure with customer data. He suggested that we stop talking about tracking customers and exploiting or leveraging their data, and pointed to a paradox – as a higher degree of individualized targeting is attempted, the greater the chance is of getting it wrong. In fact, the risk is of getting it precisely wrong – targeting customers with offers that are especially inappropriate or even offensive because of the targeting. Bad customer data, he said, citing the Hoofnagel et al research paper on user rejection of behavioral advertising, is voyeuristic, temporary, fragmented, and speculative. The alternative was to develop a proposition of genuine value to the customer, which would help build a relationship of participation.
8. Telco 2.0 Conclusions
The session tended to reinforce the notion that there is a serious opportunity here, and to give some clues as to how to address it. Sense Network’s Sibiski provided some inspiration in terms of how to link large volumes of very basic CDR data with bankable commercial objectives, and made the excellent point that there are good reasons for telcos to get started in improving their own operations and marketing practices before getting into the complexities of letting anyone else have access to the data.
However, it’s also clearer than ever that this is a challenging field and that there are serious potential pitfalls. A major, basic issue is extracting useful data from multiple databases inside the operators. Atigeo CEO Michael Sandoval made an excellent point in saying that bad targeting creates the risk of achieving negative excellence. The comparison of Google Mail and Phorm provides a good example in how consumers can respond very differently to different approaches. Finally, user feedback strongly suggests that the Customer Participation Framework is a Telco 2.0 concept that may have considerable cross-relevance in this field.
Check back tomorrow for our next installment: Consumer Services 2.0