Veikkaus' AI Strategy: Personalization and responsible gaming

We interviewed Aku Kelo, responsible for personalization and machine learning, to share how Veikkaus leverages AI to enhance the customer experience.

Aku Kelo is responsible for developing Veikkaus’ personalization capabilities, new machine learning models, and recommendation systems, as well as utilizing data to improve the customer experience. Before joining Veikkaus, Aku worked with data and AI at Posti, Pfizer, and in the financial sector.

Veikkaus serves one of the largest customer bases in Finland, operating in a highly regulated industry. How does Veikkaus leverage AI to enhance the customer experience, and what can we learn from their approach?

What kind of challenges does Veikkaus face regarding customer experience? What are the requirements for customer experience and why is it important for a company like Veikkaus?

The customer experience is a highly subjective concept. We operate Finland’s largest online store, serving over 2.5 million customers. This vast customer base brings with it many interesting opportunities.

In my experience, not all customers expect a personalized experience—it’s often a generational question.

Millennials and Generation Z, by default, expect a fully personalized experience that feels tailored specifically to them and one they can control. Older generations, on the other hand, may not be as interested in personalization. They tend to have their favorite games and preferred ways of engaging with the service.

An interesting aspect is to personalize the customer experience responsibly so that it feels exciting, relevant, and engaging for the customer while simultaneously minimizing the risk of problem gambling.

Among the online generation, it has been observed that content must quickly capture attention and remain engaging to sustain interest. People switch between services with little effort due to the ease of surfing online. This is a challenge faced by all B2C companies, from streaming services to online stores: how to provide compelling and relevant content that serves both the customer’s needs and the company’s goals?

While personalization is essential, we must also be able to cater to a wide variety of customers and needs simultaneously. In my opinion, this represents a fascinating opportunity: how to manage the overall experience in a way that ensures all parties are as satisfied as possible.

You mentioned that younger customers expect advanced personalization and entertainment. How have you at Veikkaus approached solving this, while also considering business goals and resources?

At Veikkaus, we have a matrix organization that operates across business boundaries. We aim for collaboration and setting shared goals.

For example, when launching new products, we agree between the business units on how marketing and visibility will be managed to ensure that everyone gets a significant boost and benefits, while avoiding harm to other units and products. In other words, ensuring that the new product doesn’t cannibalize existing ones.

In my role, I am responsible for Veikkaus’ overall personalization capabilities and the machine learning capabilities related to personalization. Together with the AI and analytics team, we aim to provide decision-makers with insights and analytics from A/B tests and experiments. We support the decision-making process and engage in discussions on how things can be technically implemented, always striving for open communication and transparency.

In summary, here’s how we can start collaborating:

1. Involve all stakeholders in the discussion.
2. Open communication about the current state and where you want to go.
3. Clear goals and hypotheses on how to achieve them.
4. A shared definition of success.
5. Experiments and iteration of solutions.

We try and iterate solutions until we find one that works. We involve service designers, designers, UI/UX experts, front-end developers, AI specialists, and product owners as early as possible. The executive leadership defines the goals, and our team tests hypotheses through experiments while also innovating various possibilities.

The challenge lies in balancing business goals, customer goals, and the interests of different departments. For example, the homepage usually has only one or two main personalization points, with everyone competing for visibility.

The business’s role is to find common ground on how personalization will be managed across different interests. The goal is to move toward a more holistic customer experience pipeline, where the entire experience is personalized, not just one or two personalization points. This would reduce the intense competition for the same limited personalization spots.

How have you leveraged AI in developing the digital customer experience? What concrete examples can you highlight?

We have actually been using AI extensively for personalization and responsible gaming initiatives. Veikkaus has, in my opinion, been a pioneer in digital services and AI in Finland.

Since 2011, we have been leveraging in-house AI solutions in our digital services. We have our own AI team whose role is to develop machine learning models for various needs. We also collaborate with designers and UI developers to ensure the seamless integration of machine learning models into our services.

On our website and app, we have various personalization points, where we aim to implement different recommendation and machine learning models to improve the customer experience and prevent problem gambling. If you think about our homepage, for example, there are several personalization points, each using a different model to perform various tasks.

The entire online service should consist of modular components that can be rotated and positioned in different places. We can optimize the arrangement of these modules using AI models. The content within these modules also contains various personalization points, and each point can be customized individually based on the customer’s needs, determining what content will be shown where, based on the customer’s behavior and predicted needs.

We utilize AI in many different ways: generative AI, reinforcement learning models, various offline models, optimization models, and other machine learning approaches.

Our goal is to provide customers with entertaining content. For example, on the homepage banners, we can show the most relevant or new and interesting games based on the customer’s activity history, depending on the situation. However, the most important aspect is that all of this must be done responsibly and ethically.

The next goal is to orchestrate the models so that different models work seamlessly together. For instance, we can offer lottery and scratch games, betting games, and casino games at different points along the customer journey, and the recommendation models can take into account predictions from other models, using this information to recommend games or services the customer is likely to enjoy next.

Every solution must be responsible, aligned with business goals, and solve a business problem.

What is your vision for the player experience – what kind of experience do you want to offer them?

We want players to enjoy their time with our services. We are not just a gambling company; we are an entertainment gaming company, which is why we have services like VeikkausTV, where users can watch live sports.

While gaming is important, we also want to offer other forms of entertainment and content. Our goal is for customers to find our services entertaining and relevant to their needs. However, responsibility always comes first.

We have various responsible gaming initiatives and AI solutions to minimize problem gambling. We aim to anticipate potential problem gambling behaviors and offer preventative support.

Our responsible gaming AI models detect when a player’s behavior might be becoming harmful and flag their profile. We have professionals who call and discuss the situation with the player, aiming to provide support. When someone falls into problem gambling, their customer experience becomes significantly more limited, with the goal of minimizing and hopefully eliminating the gambling problem.

Responsible gambling is a crucial issue across the gambling industry. At the same time, we are one of Finland’s largest gaming companies, and we have our own gaming studio that employs hundreds of people. We internally publish and sell a large number of various casino and e-lottery games each year, making us a surprisingly big player in Finland’s gaming scene—something that not many people are aware of. For this reason, I believe we also have an ethical responsibility toward our customers.

I believe other companies in the gaming industry should also focus on responsible gaming and ensure that players do not face financial or social harm.

This is an incredibly important issue. Veikkaus emphasizes the importance of responsible gaming and strives to prevent the harm that gambling can cause to individuals and their social and financial situations. Gambling should be entertainment, but it should be responsible entertainment.

How can companies with fewer resources than Veikkaus start leveraging AI?

A company must first ensure the 1) quality and 2) quantity of data, as AI is nothing without quality data; garbage in, garbage out. Only then can it consider how AI could be utilized in its business or life. Fortunately, the barrier to using AI is constantly lowering, and using AI tools is becoming easier every day.

One can start by experimenting with simple generative AI. Experts are also needed who understand AI and can expand its usage. Consulting can be bought externally or AI specialists can be recruited. Internal training and skill development are also important.

If money is simply not available, you need to be resourceful. This means finding time for self-study or forming useful partnerships. But the most important thing is to first ensure enthusiasm and interest, as well as making sure that data and infrastructure are in place.

AI development is currently progressing at an incredible pace. Many professionals in the field admit that they can’t keep up fully with the speed of development, so they are kind to themselves and accept that they can’t always be the first to know everything.

On an individual level, a near-term goal could be to learn prompt engineering and how to properly use AI tools. There’s no need to start building complex machine learning models, nor do you need to know how the math behind neural networks works.

It’s worthwhile to learn to use the available tools. I believe that getting “comfortable with technology” is an extremely important thing to focus on for the future.