Gaming Eminence tackles data with David Conde Head of Data & Analytics at Luckbox. Our conversation looks into the increase in data points with customers, connecting business stakeholders with data teams, types of models, and the impact AI & Machine learning will play in the future.
GE) The volume of data points are increasing at a rapid pace. What are some of the challenges iGaming/Gambling Operators face in understanding what data to use and how to leverage it to retain and grow their customer base?
DC) In this context, choosing the right information to process is key for improving data model performance, as well as avoiding noisy data that can potentially mislead business decisions.
In these circumstances, different models ranging from Fraud Detection to Marketing Retention can use very different data points, so it is important to provide the raw data and keep all the different business areas involved in this process, to avoid losing data points which are irrelevant for some of the business models, but which can be in fact quite important for others.
This issue usually happens with data points required for Fraud Detection models, which are often useless for Marketing teams.
It is also really important to have the right data in place and this is something the data providers should consider as well. For business optimisation, it is really important to have the right information, and sometimes lack of some necessary data points from data providers is another issue operators have to face.
Finally, how data is structured in order to be useful for further exploration and the own quality of the data are very critical aspects iGaming operators have to deal with.
GE) The concept of data modeling is a highly technical process and in particular circumstances not a focal point for business stakeholders. How important do you think it is for all areas of the business to be aware of how the firm's data is structured and what importance that plays in operational strategies?
DC) The key challenge to involve business stakeholders in the data structure and modeling context is to provide clear information about what kind of information can be stored, the sources that generate those data points, and the possible use cases where those data points can be used.
Based on our experience, knowing how that data is structured and stored is something that does not provide big value to high-level stakeholders, so that is why we try to focus on helping them fully understand which information available can be used for the different use cases. This means they can participate in the data modeling decisions, not from a technical side but from a business-oriented one.
Finally, it is important to highlight that there is no way that the any technical data team could create great value for the business without getting involved business stakeholders who can bring business expertise into the table. For any great data model both technical teams and business experts should work together there is no other way to success.
GE) Typically there are three main types of data models used within organisations - Conceptual, Logical and Physical. As an iGaming/Gambling Operator, what purpose and advantages do you feel that each serves in assisting with business growth?
DC) Conceptual models aim to provide common name convention and high-level data points for business stakeholders. They are mainly used as a way to facilitate non-technical colleagues with understanding of the data models.
Logical models provide a high level of abstraction from fine-granular data points details. They are mainly used as an interface between very granular details in database tables and business monitoring. This can be useful in KPIs monitoring, where each department can make use of these intermediate data models in order to build their own dashboards without depending on the data analytics team.
Physical models are the models which represent really how the business information is stored in the different databases. They are crucial, especially in the igaming industry due to the huge amount of data and its nature, thus, optimisation of those models is critical. Storing only the data we really need for the different models and designed the data infrastructure in the optimal way are really necessary in igaming business
GE) In creating data models at any organisational level, data quality can be critical in the ability to leverage data correctly. How significant are the risks of bad data quality for operators?
DC) Data quality validation is a must before any kind of data model can be implemented. The risks of underestimating this issue are significant - including rewarding, rather than limiting, bonus abusers; failing to detect fraudulent activity; accidentally duplicating promotions, offering misaligned campaigns, or missing retention actions. Having the right data with the right quality is key for business optimisation.
GE) For operators that haven’t looked at exploring AI and Machine Learning, what are the potential benefits, and what data framework needs to be in place to ensure this technology is utilised correctly?
DC) Within the face-paced world of igaming, new approaches such as AI and machine learning are essential if operators really want to scale their business.
New customers can be managed differently, with more relevant marketing campaigns for cross sell or retention based on better performance, cross sell or churn detection models.
Customer operations can be more personalised based on betting recommendations via Machine Learning algorithms and bonus abuse and fraud can be more effectively combated thanks to AI data models.
Those machine learning algorithms not only provide higher escalation capabilities to the business based on process automation, but also provide the right flexibility and adaptation.
This is key for moving from a static and well defined customer base to very dynamic circumstances, where any customer behaviour is different from the rest and it is there where artificial intelligence algorithms are really successful because the approach is based on learning from the data, rather than learning from pre-fixed business rules.
Regarding the kind of data framework which should be in place, for machine learning data models raw data should be available, the quality of the data should be validated and at this stage it is really key to have a considerable amount of historical data so that those algorithms can learn from the data.
About our contributor
David Conde is Head of Data at Luckbox, a betting platform built for a new generation of players, which offers wagering on esports, sports and casino games. David, formerly of Ericsson and online gambling companies MoPlay and Gaming1, leads Luckbox’s business intelligence infrastructure, which underpins the company’s data-driven approach across multiple departments
Luckbox is an award-winning betting company that offers legal, real-money betting, live streams, and statistics on all major esports and sports as well as casino games on desktop and mobile devices. The Company has a Business-to-Consumer (B2C) platform, and by leveraging shared technology, data, and resources, the Company can offer an extensive range of betting options for esports tournaments. Visit https://luckbox.com/