Offloading of data to big data platforms
It is not even that long ago, when a new competency focusing on Big Data emerged at Adastra. After a series of subprojects, the market is now ripe and the company is gradually preparing to comprehensively take on “big data” and maximize their contribution to their own business. How does Jakub Augustín, the new Big Data Competency Leader, see this challenge and how should the market and companies prepare for it in the coming months?
After four years of working at Adastra you have now become the new Big Data Competency Leader. Is there something new for you in this position? What preceded it?
It’s actually not that new to me. Roughly two years ago, when I was a consultant in the banking division, the recession also hit us, like it did to number of other companies. There was less work and it was clear that something had to change. It was at that time that Adastra decided, which I still appreciate, we had start doing new things and enter new markets in order to reduce the situation. So I actually got a chance to work as a consultant on a project in Russia…
I wouldn’t have personally chosen Russia as a work destination, but as was proven later, it was a great experience. We worked at the client on a project related to data warehousing, which is one of the core competencies of Adastra. But while the majority of Czech banks had a data warehouse at the time, the Russian market was behind us. I got the opportunity to work on a complex project from the start. I remember a funny situation when we arrived at the client to carry out an audit of data warehouses and it soon became apparent that there were actually no data warehouses yet… but in the end the project went well and the client was satisfied. It brought me many new things and reassured me that the expertise of professionals from Adastra is truly world class.
Upon my return to the Czech Republic the recession still could be felt. Big Data was then a novelty. Still some two years back it was not yet established on the Czech market. In any event, it was an interesting area of potential. So, I began to work on it alongside my normal workload. From the beginning it was rather a “one man show”, but then it began to grow and new people were added and independent Big Data competencies appeared at Adastra with their own projects. So, my position gradually evolved.
I take it as a great challenge and commitment. In order that we could work with Big Data, other colleagues have to continue taking care of the projects that have been and are crucial for Adastra. We must prove that Big Data is not just a marketing “buzzword”, but there is a real business behind it. In addition, now is the time that Big Data has become fully applicable in practice. Until now, everyone was talking about it, included it in smaller projects and studies, but companies did not fully commit themselves to it yet. Currently, we have finished the introductory phase at some of our clients and are gradually implementing their own Big Data technologies.
Within the framework of Big Data we work with such data sources that the bank itself can not properly grasp and exploit, because they are not only bulky, but also to quickly created or unstructured.
Let me give you another example concerning the analysis of shopping baskets. We’ve known it for some ten years, but so far it has never been done because of the high costs. Thanks to the new technologies available, it is finally possible. Moreover, we are also able to move the reaction time (for example in product offerings) from months to days or real time to quickly respond to current life events and client needs – for example, the birth of a child or an apartment renovation. This is called Event Driven Marketing.
Gradually, there will be more common methods for concepts such as text analytics, predictive modeling or geo-analysis in connection with Big Data, which is also our goal.
Marketing is certainly one of the key areas. But we would like to move it more from selling products to providing services, where there is a reaction to life events in real time, whether we find the necessary information from text, voice or data activity on Facebook. This is then related to better options for micro-segmentation as well. Another area of use is also risk management, for example.
There are individual technological solutions alongside this. I see it with our clients who really have more data than many. For example, like when the Telco segment is transitioning to a 4G network it generates a larger amount of information which increases the number of technological challenges, as this data is not processed. There is a lot of hidden potential in the technology itself, you just have to just technologically take it.
For example, we recently completed a project on text analytics, based on which we mentioned the concept of event-driven marketing and built a completely new marketing concept for one of the largest domestic banks. To this day, the company routinely practices “carpet bombing”, which they distribute to thousands of clients asking if by chance they do not want a credit card. Our project is specifically focused on specific niches, including people who are about to leave on holiday, which can be read from the data and target an offer to them with suitable products. We managed to raise response to the campaign multiple times, which is great.
Before that we were working on more interesting things, when we predicted the age and sex of users who clicked on a Web log.
Yes it is. If we have a sample of the users themselves, if stated for example that women aged 30 (maybe 2%) and we knew nothing about the rest, we can deduce according to certain patterns of who it is in the second case. As a result, we have got so far as 80% of identified users, we were able to put a high probability according to who clicked where.
Analyzing visual data is currently one of the hardest areas being implemented. For example, analyzing audio data is much easier. In any case it works similarly to a Web log. This is done through machine learning, where the machine compares certain similarities using certain pattern similarities of characters, and includes the image in a particular category. However, it cannot identify a specific entity, but its features. For example, we have defined the characteristics of a cat, which you are able to compare the characteristics of the images and mark resemblances. The areas that you can focus on are countless and it is necessary to choose which direction to go.
2016 will be a sign that in general all clients begin using Big Data technology themselves. These changes will have to happen first, so that they can develop their own projects and also shorten the reaction time. Companies probably also still do not want to use data external parties, but start on their own. Banking will move more toward event-driven marketing. Mobilizing multiple electronic channels, which in my opinion will gradually reduce working branches, where the clients leave a huge amount of information, but the staff is not able to handle such an amount. The customer already does not necessarily have to come to the branch, but the bank will still be able to offer advantageous products remotely.
Unfortunately, there will be, as we are changing the way of working with these technologies. Until now, when a client had chosen a technology it wasn’t possible to be confused about it afterwards, because the investment was too great. New technologies are so-called “open source” and allow you to choose from multiple solutions. But I don’t think that the confusion will last. The direction is fairly clear, which is good.
I always tell clients: “Choose areas that you are worried about, and we will make business process outputs in a short time. There’s no sense in starting with technologies if you don’t know the purpose of the whole work. Have a few suggestions created that basically cost you nothing (the benefits of open source) and decide.” That's the process I would generally recommend to anyone who wants to use Big Data in their business to begin to solve complex tasks.
He has been working at Adastra since 2011, when he joined as a consultant in the banking division. He gradually began to build a new Big Data competence roughly from 2013, which is currently the market leader. He has extensive experience with data-oriented projects for major clients from the banking sector in the areas of Data Warehousing, Master Data Management, Data Migration and Data Governance. He has also worked as a consultant for a client in Russia as part of a complete redesign of a data warehouse.