If your big data analytics project has struggled for resources and backing, then a solution launched back in June by Wipro Limited could be one option to unblock it. Data Discovery Platform (DDP) is a packaged response to the problems which many organisations have struggled with when trying to get traction for next generation analytics - legacy IT, lack of analytical capability, high entry costs.
When DataIQ spoke to the company about its new proposition in early Autumn, it already had 25 live clients onboard. But even more impressive are the behind-the-scenes resourcing and architecture which have gone into DDP.
Pallab Deb, VP and global lead for analytics at Wipro Limited, outlined the path that had led to the launch. “In our business, we need to see what the future holds. Our analytics business now is very large - one third of our revenue at around $1 billion. We also have a lot of business in traditional areas of managing large data assets for clients, driving out reports, cleaning data, supporting business decision making from those reports.”
Deb noted, however, that, “things have chenged over the last few years. Our clients are facing a situation where getting reports from data sets is no longer enough. They are now looking to make use of data from outside the enterprise, like social media, to understand business drivers for their partners and external stakeholders. It is no longer enough just to look at data from within the enterprise.”
Two examples demonstrate both the new analytics requirements in the market and also how DDP is set to grow through a “land and expand” strategy. One financial services company based in the Australia-New Zealand region wanted to use a broader set of data and run marketing analytics quicker. Its existing model used a data warehouse based in the IT department and had a lengthy turnaround time for new data sources.
Using DDP, the client was able to accelerate its time to market, generate more predictive insights, be more flexible with the data it included and develop more models. It was also able to include external data, such as economic indicators, into its customer attrition models.
Jayant Prabhu, global practice head, big data analytics, for Wipro Limited, explained: “That opened a big window because the client didn’t just want to look at what the customer was doing, but also external factors and what was happening in the region. That was not possible in their traditional approach. That is now a long-term contract and we are their platform of choice for insight and analytics.”
In another example, a US-based utility was concerned about its customer experience and wanted to reduce costs, but was limited in what it was able to change because of regulations. Using DDP, it was able to solve the issue of peak voltage consumption using smart meter data by segmenting customers and identifying groups who could be made offers to reduce their energy usage and be encouraged to run appliances at low consumption periods.
Together, these examples demonstrate the breadth of big data analytics tasks being undertaken and also the reason why existing infrastructure and resources are struggling to cope. Traditional data warehouses are typically not architected to accept external data flows and IT is often averse to allowing predictive analytics queries to run on core business systems for fear they will affect critical performance. Few have been set up to deal with data streams from the internet of things, either.
Deb says that technology is one of the barriers to analytics adoption which DDP has been designed to overcome. “There is also a lot of innovation being led by open source technologies which have been scaled out for big data and have become relevant to enterprises which are more used to Oracle, SAP and IBM. There has also been a lot of acquisition and merger activity within the technology space,” he said.
“Our point of view is that it is important to be open to this new world of data technology, so we took a look at what’s available and how to put together a soluton that was replicable, available to multiple customers, affordable and easy to deploy, as you would expect from new technologies.
We have started on that journey and how to make technology easier to adopt, as well as adding some ‘magic sauce’, so what we do is not easily replicated by our competition,” Deb added.
As well as using solutions that originated in open source, such as Apache, Hortonworks and Cloudera, Wipro Limited has built a suite of analytics apps to sit over the top of the data management piece. He notes: “If an organisations has already got a data lake, great - if not, we can help them to create, prepare and integrate data in an automated environment. Once that has been created, we have got hundreds of analytical apps and data models pre-built, so if the challenge is customer churn or price optimisation, clients don’t have to build that from scratch. This way, they willl get value out of data quicker.”
The second major piece of resource being offered via DDP is a huge data science team of more than one thousand analysts, based offshore in Bangalore and soon to be augmented with an onshore group. “We’ve focused on building that capability, scaling out that resource,” says Prabhu. “We have focused on becoming a vital element further up the value chain, understanding the value of analytics to our clients.”
He adds: “Technology has not necessarily got easier - we still do that - but getting the right insight is the critical space. The data discovery platform reduces the time to market. That is an area where organisations have struggled to get insight in time and understand their next action. It is aligned around the business questions which stakeholders will have.”
“Typically, for customer retention marketing, an analyst will create a regression model and random forest model, but a marketer has no idea what that means. Often, clients don’t know what questions to ask, so we can help stakeholders to understand what they should do. Where our data science teams come into play is building applications and running them. It accelerates the time to market and anaytics adoption rate because there is a straight line to ROI,” says Prabhu.
For Wipro Limited as much as for its clients, building analytics-as-a-service has involved unlearning some traditional ways of thinking. It has also demanded a lot of effort ensuring its data scientists have rounded skills, beyond just mathematics and coding. To do this, the company has created bespoke programmes in partnership with universities in the US and India which combine machine learning, statistics and coding with interpretation, story telling and domain understanding. He explains: “Data scientists need to be able to communicate to key roles in industry, like CFO, CRO, COO, etc. If you go to an academic institute, they teach you the theory, but the application of it is often not up to scratch.”
Another innovation in DDP is its pricing model - simple, medium and complex, each including a number of models and apps which subscribers can use. Says Deb: “So far, clients love it because it is very clear. They can create insight, refresh it regularly and run analytics across the year and know the cost.”
Traditional IT systems and legacy architecture are set to continue for some years to come. But alongside them, a new generation of big data management and analytics architecture is emerging, most of it cloud-based and increasingly delivered as-a-service. Wipro Limited’s DDP is a manifestation of this trend - and a heavily-resourced one at that.