Thought Leaders in Big Data: Interview with Prem Uppaluru, Co-Founder, CEO and President of Transera (Part 4)

Sramana Mitra: So you have the ability to do routing at a granular level.

Prem Uppaluru: Exactly. This brings me to the next use case, which concerns the largest direct response marketing merchant with multiple ways to attract customers. They are the company behind the [skin care product] vendor Proactiv. They get around two million calls per month from customers responding to their TV, radio, print, and web ads. Customers call in to order different brands. We manage all of those calls, both for sales and for service. In that case, the sales calls go to a different set of vendors, who take those calls and do the order fulfillment. We measure their performance at the vendor level and prioritize one vendor over another, sending the calls to the highest performing vendor because that is the level at which the vendor is given a scorecard. With this scorecard, we can score agents or agent groups and route high-preference calls to higher performing agents.

SM: This feels like a data analytics problem, but it is not a huge amount of data you are working with. Routing to different vendors doesn’t seem like a very complex data issue. There are a few of these vendors. People don’t work with big data numbers of vendors. Where is the big data angle to this?

PU: The data of interactions – the millions of calls. You have to measure each vendor based on what it is doing for each call. Each call is an opportunity to convert, just like a web hit. A web hit either converts or it doesn’t, so you have to start doing analytics. It is like saying, “Yahoo only has one website, so where is the big data?” It comes from the fact that Yahoo has millions of visitors.

SM: Routing two million calls to a relatively small set of vendors is not a big data problem.

PU: I beg to differ. Big data has been used for many things. From our perspective, big data is not about the volume of data. It is about the velocity, variety, and complexity of the data. That is where we find the value – accessing that in real-time and allowing real-time decision making.

SM: What you are doing is extremely valuable – I don’t contest that. Nowadays the strategy of everybody is to label everything as big data because big data is hot and trendy. I have a bit different view on what big data is.

PU: We do not view big data in terms of petabytes, but we do view it in terms of terabytes. The real value is the NoSQL aspect of big data. That is what we exploit, because some of the customer interaction data that comes from multiple operational systems is structured, and some is unstructured. A chat session or a speech, for example, are unstructured, as are weblogs. Integrating all of that in a single database on which you can make queries to draw segmentation and profile customers and the outcomes of interactions – that is where we see the value of big data.

We do have a Hadoop cluster. It is not a huge cluster from a data perspective, but it is extremely useful from a cloud delivery perspective. We have a fairly inexpensive data repository into which we can pump data that is unstructured, semi-structured, and structured, and pull it all together to do the analysis, pull insights, and make routing decisions. That is what we use big data for.

This segment is part 4 in the series : Thought Leaders in Big Data: Interview with Prem Uppaluru, Co-Founder, CEO and President of Transera
1 2 3 4 5