Better and faster intelligence, to more people
To stay ahead of the competition and stimulate business, Companhia de Seguros Tranquilidade needed a way out of its over complex and slow-moving legacy data architecture to enhance the speed, transparency and access of its reporting and to dig deeper into the possibilities of analytics.
The insurance company Companhia de Seguros Tranquilidade was founded in 1871 and has, for a long time now, played a leading role in the Portuguese economy. Its physical distribution network encompasses 38 own stores, more than 78 franchise stores and over 270 agent stores nationwide. In 2011, its 770 employees offered Tranquilidade’s full range of non-life solutions to their 618,000 active clients.
As is often the case in information-rich and regulation-bound industries like theirs, Tranquilidade was struggling with the severe effects of data explosion on their business agility. Not only were they getting close to a point at which their data warehouse would no longer be able to hold all of the fast-growing information, their reporting and analytics processes were forever increasing in complexity and decreasing in speed, while the delivered documents were getting more difficult to interpret.
The reason: from a legacy point of view, they were burdened with complexly layered architecture in which the data were stored in SAS files and spread over 8 separate datamarts of 25 terabytes, each dedicated to a different corporate department. Reporting and analytics could only be handled by highly technical experts by means of aggregation, which was accompanied with the usual drawbacks like data deduplication and scattering of intelligence. It was high time for a change.
On a technological level, Tranquilidade selected the Greenplum database because of its scalable shared-nothing MPP (massive parallel processing) architecture. “We strategically opted for this kind of structure because it enables us to scale horizontally more easily and stimulates faster analytics than the more common OLTP anatomy of the solutions of the competition”, explains José Manuel Vera, CIO of Tranquilidade.
But the two main arguments why Tranquilidade chose the Greenplum database are as straightforward as they are vital: speed and transparency.
“Fed up with our labyrinth-like data architecture, we were immediately attracted to the crystal-clear model and infrastructure of the Greenplum database that is built to manage, store and support analytics of massive amounts of data”, said Vera. “It allows us to consolidate all the raw data into one single data warehouse and have all the separate data marts communicating.” This has, of course, huge consequences for both the speed and complexity of the reporting and analytics.
More accuracy, faster delivery
“Our former data architecture used to be so complicated that only advanced technical profiles were able to understand it and turn the data into any kind of intelligence, even for the most basic kind of querying and reporting. Not only did this influence the delivery speed but it left the door open for mistakes. Worse, the reports that were made, were often so intricate that is was nearly impossible for untrained business users to interpret them correctly.”
Those days are fortunately over. Thanks to the straightforwardness of the Greenplum database, Tranquilidade is now training its users so that there will be department-wide access to the reports and employees will be able handle their own queries. The most savvy data managers and the more technical profiles are already up to speed and now representatives from the sales and marketing departments are following the course as well.
“There is no point in having powerful tools, if the information is inaccessible to the business users”, said Vera. “The more people who receive the reports and intelligence at the right time, the more we can use these to inspire successful decisions and follow the right strategies.” That is why Tranquilidade is working on giving their external partners (near)-real-time entry to relevant reports.” While for some bigger partners this means ‘just’ the speeding up of the process, other smaller agents, like repair shops, now receive opportune information where before they did not. This proves how a ‘simple’ shift in architecture can even end up raising the efficiency of external partners.
Complex reports finished in 1 hour, instead of … 48
Tranquilidade was aiming for an average increase in the query speed by tenfold and is very satisfied with the results driven by the Greenplum database. Monthly reports that used to take 48 hours of straight processing now run in less than an hour. “The speed at which we now can upload information is incomparable to anything I have known before”, Vera added.
“On the basic level of reporting, we were enabled to boost the delivery time from a monthly to a daily basis. Knowing each day, for example, how many policies we are selling and differentiating between the separate divisions has, of course, made a huge difference for the commercial department”, Vera explained.
“As to the less prioritized and more technical reporting, for regulators and internal use, we have decided to precipitate from once a month to once a week”, he added. “This is the kind of information that, for instance, gives us a much better perception on the contracts we have with all the other companies that work in the claims management segment.”
Defying the rule of pi
Even the process of moving the data from the old to the new data architecture of the Greenplum database – a very complex and elaborate operation – which was calculated to last up to three years, has demonstrated amazing velocity. “The process was started at the end of 2011 and we are expecting it to be completed before the end of this year. So it took three times less than foreseen, which is ironically the exact opposite of the rule of pi, which states that you should always multiply the estimated time for IT projects by 3.14.”
Tapping into to the Goldmine of Big Data
“One of the main benefits of the Greenplum database is that it allows us to move much deeper into analytics and even tap into the goldmine of Big Data”, said Vera. “Besides asking better questions and organizing more complex ETL processes, we can now take full advantage of predictive analytics.”
The Greenplum database facilitates high performance parallel import and export of compressed and uncompressed data from Hadoop clusters. “So not only are our structured data now better structured, we can also integrate them with unstructured data from Hadoop sources, like clickstreams, to extract deeper insight”, said Vera. “Our direct line insurance company LOGO was, for instance, able to find out why there was such a discrepancy between the amount of online simulations and actually concluded contracts and successfully adjust its strategy in the matter.”
Giving the business what its needs
“When it comes to basic intelligence, we are now spending a lot less per report, in man-hours, and because we are able to do a great many more than before”, Vera concluded. “And, more importantly, the business collaborators now receive all the intelligence they need at the right time so that they can focus even more efficiently on their core activities and drive revenue.”