Improving our customer’s experience and our operational efficiency
By Lonnie Hamm, Kindred Group Head of Data Science
It is not news that the amount of data being generated and stored in the world is increasing at a fearsome pace. This pace is expected to grow even faster as output from more processes and devices are digitized.
For many, or perhaps most companies, the amount of data they are collecting has outpaced their ability to do anything useful with it. The challenges are three fold: culture, technology, and people.
As a purely online business, Kindred generates an enormous amount of data every day. In 2016 we averaged 16 million transaction per day resulting in about 6 billion monetary transactions during the year. This is similar to what PayPal had in 2016. Several years ago, Kindred put a strategic focus on being able to better exploit this data, with a goal of improving our customer’s experience and improving operational efficiently. In this blog we’ll provide an overview of our cutting edge data and analytics capability. We’ll do this in the context of the culture, technology and people challenges mentioned above by considering in turn Kindred’s Data Culture, Data Technology, and Data Science.
A company with a data driven culture is one where decisions, when possible, are backed up with data analysis and where outcomes are measured. A data driven culture will leverage and multiply the effectiveness of investments made in data technology and data science. The lack of a data driven decision process leaves one’s data underutilized.
For Kindred, data has always been central to the way the organization is run and strategically, long term management initiatives have put data at the centre of the strategy. In fact, organizationally, the data analytics groups sit under Strategy. Our vision is to make Kindred the “Most scientific, data-driven gambling company”.
At the core of a strong data culture is a strong technology culture. Kindred’s technology culture is the backdrop against which we were able to build and deploy our cutting edge data and analytics technology. A few examples of our technology have been written about in the previous blog posts “230 transactions per second…” and “From Monoliths to Stateless Microservices”.
The volume, velocity, and variety of data are all increasing. This creates challenges in storing and using data effectively. To meet these challenges, Kindred built a big data platform to supplement the existing Oracle data warehouse. Our big data platform is built around Hadoop with other associated components such as Cassandra and HBase. The scalability of this platform means we can cheaply scale the amount of data we store and process by simply adding nodes to the cluster. We can think of the data in our big data infrastructure as being “data at rest” or “data in motion”.
The data in motion is data that is being processed in real-time. A key objective in building the new data platform was to ensure that it was part of our “live production” environment. Building processes on the big data platform using technologies such as Kafka, Spark Streaming, and Storm enables us to take real time action from our data insights and deliver increased value to our business and customers.
The data at rest is historical data on which we can do research and build machine learning models on. Hadoop gives us the ability to store data across many different formats, e.g. JSON, XML, RDS, CSV, and others. This flexibility in format means we can perform queries across structured, unstructured, semi-structured, and poly-structured data. This is something that is difficult, if not impossible in a relational database model. Furthermore, using technologies such as Spark and Hive, we can perform these queries on huge amounts of data. Queries that aren’t feasible to perform using SQL in a traditional database.
Given our ability to process large amounts of varied data types, conceptually we are only limited by our imagination in the ways we can use our data to improve the customer experience and improve our operational efficiency. However, there is a human constraint in the number and quality of technologist we can employ to maintain our data technology ecosystem and the number of data analysts to do the data science work.
Data and analytics isn’t anything new at Kindred. What we have done in recent years is successfully added teams and advanced technologies to greatly extend our capabilities. The success of this effort is due in part to the fact that we weren’t trying to build these new capabilities from scratch or from a small base.
The teams added in recent years are big data, web analytics, and data science. These new teams are part of a wider long standing data and analytics department consisting of 70+ people. Collaboration across these departments and within teams is strong with cross-functional teams frequently working on the same projects.
The Data Science team‘s remit is to use machine learning and A.I. to tackle some of our biggest problems. They have a range of skills and backgrounds holding higher degrees in disciplines such as engineering, math, statistics, and economics, all with strong domain expertise. Tools and languages the team commonly use include R, Python, Scala, Shiny, D3, Hive and Spark.
Kindred’s big data team are responsible for managing our big data related technologies and the data within. They also build and deploy production standard solutions into our live production environment from proto-type solutions developed by the data science team. This strongly functioning end to end collaboration between the big data and data science teams is a real strength for Kindred. Some organizations struggle to implement solutions even if they manage to build some data science capability.
The big data team also works on initiatives for the larger Kindred organization. For example, enabling certain parts of the business to respond in a more real-time manner to customers. This democratizes the technologies across all parts of organization and leverages our capabilities.
Data as a Differentiator?
At Kindred, we’ve set ourselves the goal to be innovators in the way we use data to make decisions and to supply an industry leading customer experience. There are millions of decisions taken every year at Kindred. Small gains on some of these translates into big gains in efficiency and profitability. Huge opportunities also exist to revolutionize the customer experience through innovative use of AI and data.
Numerous projects have been implemented or are in process since these investments in big data and data science began, adding considerable value to our business. Some examples include churn prediction, high value customer identification, fraud detection, onsite personalisation, improvements in player safety and AML processes, understanding behavioural drivers – emotions, sentiment and motivations of customers.
In future blog posts, we’ll expound upon our technologies and some of our major initiatives. But, in the meantime, if you believe these are areas where you or your company can help the Kindred Group innovate, or you are a data scientist or big data technologist, we’d be very interested in hearing from you by contacting us at firstname.lastname@example.org