Last week in Amsterdam, the TM Forum’s Big Data summit caused a few old (by which I mean experienced) telecoms professionals to change their minds on the subject. The consensus is that it is at the top of the hype cycle and about to plunge into the ravine of reality. And that – to me anyway – means that it is time to concentrate on the reality of it, as it relates to telcos.
Amongst an interesting line up of presentations, (and there are excellent summaries of the event from Tony Poulos here and Light Reading’s Ray le Maistre here) three caught my attention. A senior architect at eBay certainly made it plain that they are a company that deals with big data. Their biggest table in a database holds three and a half trillion lines and they deal with 150 terabytes of data a day. Having proved that we are indeed talking ‘big’, one of the questions from the floor then had me seriously re-considering the idea that telcos have some kind of golden relationship with customers. The question was to what extent eBay used data from CSPs to better understand their customers. The speaker simply did not understand the question. When he realised that a CSP was, for instance, an internet provider, he laughed. And then basically said that they had quite enough data on their customers to be going on with, thank you so much for offering. An interesting side bar was that they had discovered that very few customers actually type in search terms that take them straight to the right thing – these are normally two or degrees of separation away. This means that their ability to offer intuitive suggestions is paramount to their business and a tangential but important point for telcos.
Another presentation that had us all thinking was from Kate Craig-Wood, founder of Memset. She said that although data might very well be big, a lot of the data today is also quite ‘thin’ and therefore big data techniques are only required in a relatively small amount of cases. (To confirm: she actually said that at an actual Big Data conference to actual big data people). Take the graphics and videos out of Facebook, for instance, and you have text and actually not very much of it, relatively. Same with Twitter, there may be billions of tweets a day but it is still essentially text. Her thesis was that the whole theme of big data had surfaced because SQL does not scale. She is convinced that this problem will go away and so will the current challenges of ‘big data.’ There are, she said, many brilliant minds working on the problem. She said this, I thought, with one of those knowing smiles.
Privacy was the subject that brought us back to a telco’s role in the current environment. Using all this data is fine in theory but will customers react and pull the shutters down on companies who misjudge how far and how fast they can go in offering tailored services to customers and cause the customers to unsubscribe. The trick, said the speaker from Boston Consulting Group, is to go at the right speed. Gently, cautiously, ask customers if you can use their data to offer them more relevant services and slowly customers will trust you.
And trust is where telcos need to play. As we have said before, the service provider’s role is to be the trusted access partner, the identity broker for their customers as they consume services from wherever they choose. In the meantime, telcos can use their own ‘big’ data analytics techniques in other ways – Bell Canada for instance, helps OTT partners better target the five billion adverts that are served over their network, thus making money out of services that in themselves make them no money at all.
Next time: the human stuff that will make your Big Data projects succeed or fail.