SBC’s Jessica Welman, editor, SBC Americas, recently joined IGT Vice President of iLottery Products and Services Karri Paavilainen and IGT Senior Director for Data and Advanced Analytics Alan Mackey to discuss leveraging the data revolution for lotteries.
SBC: Let’s start with the broad concept of data and how it relates to lottery. I think to an average consumer, it still feels largely — at least in the U.S. — an anonymous process when you purchase a lottery ticket. So, can you please correct what is wrong with that assumption?
Karri Paavilainen: About 90% of lottery play is still anonymous, even in the states in the U.S. that have an iLottery program. But the good news is, this is now in a transitional phase. Many lotteries have initiatives in process to improve the management and use of player data. These are mostly limited to a single dimension of the data – not unified across the different platforms that lotteries have.
Alan Mackey: From my perspective, the lottery industry has always worked with data very well. Lotteries’ systems can handle large volumes of data. What is happening now is lotteries are transitioning towards digital platforms with iLottery and omnichannel options. Traditionally, the host system supplied lotteries with most of their data. With iLottery, data is supplied from many different systems, and lotteries need to harness it all so it can be used effectively. It’s a big transition, an exciting time and opportunity for lotteries, as they move towards better understanding players and being able to use that data to gain insights that previously were not available.
KP: There are also three key challenges that lotteries must overcome. The first is structural: how to integrate data from different systems, from different vendors. There’s a retail system; there’s a frontend, a mobile app; there could be an iLottery platform, loyalty program, and second chance. Different platforms lead to situations where there’s no single repository to integrate all this data that could be available from these existing systems. The second challenge is that all the data is in different formats. There are different identifiers for players which heavily limit the usability of that data. It requires cleaning up, organizing that data, and pulling it to a single spot. And the third challenge is the availability of data. Most of the retail player data today is still research-based. The key source of analysis for most lotteries is the physical retail location, not the player. This is a natural outcome because of the limited digital data touchpoints to capture the player-level data.
SBC: With the need to synergize data, how do you decide what the most important data sets are? How do you identify top-quality data?
AM: Data needs to be correct and complete so that people can make thoughtful decisions and do effective analysis. Core to everything is the player, and then having player data from multiple sources. Data comes from everywhere, but it needs to be harnessed and transformed into one system around certain things: the player and lottery operations. Data is quite varied but needs to be as complete as possible.
KP: The essence of big data, machine learning, and leveraging AI is that you have large data sets, and the key question is how to pull that together in meaningful data points, with meaningful analysis. There’s data scattered across different platforms, transactions, and website visits. What products did the players see there? What winning numbers were they reviewing? How often did they do that? The more data we have, the better we can convert that information to actionable insights; for example, ensuring that players would see more relevant advertising and offers.
SBC: On that platform note, is it Cloud-based? Is that how IGT houses this data?
AM: The Cloud provides us with the opportunity to store as much data as we can in a very cost-effective way. It also provides us with the ability to process and analyze that data. With traditional, on-premise systems, we may have been limited in the amount of processing we had, or the amount of storage. The Cloud enables us to process a lot of data and take advantage of AI and/or machine learning. These algorithms are very hungry as far as processing power is concerned and update themselves, so Cloud technology allows us to do that, in a scalable way. Over time, data builds up, and you will need scale with your platform. Another thing Cloud offers, which is less talked about, is innovation. Most of today’s emerging innovations are coming from companies that are involved in the Cloud – like Amazon Web Services (AWS) and Microsoft. The Cloud affords us the ability to leverage innovation very quickly, in a secure way. The Cloud also has many other services we can take advantage of in terms of backup and integration with other systems and has been revolutionary for data.
SBC: Does deploying systems to the Cloud make it easier for your customers?
AM: Absolutely. What would have taken weeks or months to provision before, in terms of hardware, we can do in a matter of minutes or hours. The deployment of these systems is very rapid.
KP: Updating systems is vastly improved, thanks to Cloud technology. When you have a non-Cloud solution, the key challenge is how to get the latest product updates deployed in a relevant period. The Cloud solution gives lotteries the ability to “plug and play” the new features that IGT is developing. That’s a key factor for futureproofing a lottery platform, that you can plug and play new features as they come to market.
SBC: Thinking about futureproofing, we have heard a lot about internet browsers changing third-party cookie policies, and it can become difficult to aggregate data profiles. Why is it important to have a singular source of data around a player that gives you a very robust picture? Is that in part because you are trying to futureproof to changes in how the Web works?
KP: The idea about player data is not to be dependent on third-party data, which most of the advertising platforms are. Lotteries can leverage player data in their different marketing initiatives, whether they are driving marketing through paid media platforms or through their own platforms and customer relationship management tools.
AM: For data, from the accuracy and consistency point of view, it means a lot from a compliance standpoint, from security, for improved customer experience. We need accurate data about a player, and it must be centralized in one place. Especially with AI and all the exciting things happening now. AI relies on data, and if it’s not correct, if it’s not centralized, then you’re going to get less than optimal results. To use accurate data and not put your lottery at risk, you need to have one accurate player profile. Effective decision making, even for human-level decisions, requires accurate data. In terms of an offer model, if lotteries do not have correct data, then they’re not going to be able to make the best offer. It is essential that you have one single source of truth for the player.
SBC: IGT has a Player Data Platform. How is this offering shaping what data your operators and partners are looking at and what they can do with it?
KP: The true power of IGT’s Player Data Platform (PDP) is twofold: to build player relationships, and to make sure that the products lotteries offer best match player needs.
AM: We also have a churn prediction or churn prevention model which measures the level of engagement. We can create sustainable players with greater lifetime contributions by having visibility into their patterns of engagement and purchase. We can see their preferences for games, and trends with those games. In terms of channel, not all players are online players; they may prefer the retail channel and we can identify that accurately based on how they use digital services. Promotional effectiveness and the best time of the day can be measured. Another huge benefit is our sophisticated recommendation engine. We can very accurately offer games that will interest players, making the overall player experience better. All those benefits are made possible by using data and advanced analytics.
SBC: You mentioned that previously, surveys and soliciting feedback from people after they had done something was the source of data that lotteries were working with. Are lotteries surprised to realize how much data they have now at their disposal?
AM: Consumers don’t want to go to websites and search for things. It’s amazing this has happened, from people wanting to remain anonymous, and fearing not having privacy, to being very much open to providing data to get a more personalized, better customer experience. It has become essential, as with other industries, that we move ahead and provide these personalized services to customers for their players.
KP: We need to first offer digital services to players to capture more of those data points. IGT has an integrated retail lottery and iLottery system helping us create the omnichannel experience for players. Lotteries can offer mobile apps, digital play slips, digital tickets, ticket scanners so players can check their tickets, and digital loyalty programs. When lotteries make these digital touchpoints and services available to players, we capture more of the player-level transactional and frequency data.
SBC: What are some examples of how the consumer experience can differ on the personalization level when they are engaging with the product?
AM: That’s where the recommendations come in. That’s where propensity modeling comes in, where we can understand what games will engage a player or what they would like to play or what sort of experience they would like to have on the website. Personalization comes in many different forms, including with generative AI now coming to the fore, which is going to have a significant impact. We also can change messaging to players to make it most relevant to them as well. How we treat players can get increasingly personal — from how we form the relationship, to how we communicate with players and how we offer a more personalized experience for them.
KP: User experience is the next level of personalization that is coming. Currently, a player may see a certain promotional offer, but we are still looking deeper at how to organize the front end so that it serves the player’s individual needs. Something we continue to work on is creating personalized user interfaces with features and functionalities relevant to the specific player. We are also currently focused on ways to communicate more effectively – the content, the timing, the channel. The next level is to improve the player experience, or even to totally transform how the user player experience is constructed.
SBC: Why do you think lotteries should join the data revolution? What should their expectations be, and what are reasonable goals to set for themselves by engaging more actively with the IGT PDP data platform?
KP: The greatest successes over the years have been achieved when both the vendor and the lottery are willing to look at the business, how to improve it, and how to build it, together. And then obviously a vendor partner needs to be capable in technology. They should preferably offer an omnichannel integrated solution because that really creates a full lottery view of a player, not just the digital channel as separate or retail channel as separate. And then the vendor must be able to offer services to convert the data into insights. It’s not just the technologies; it is also the services, the people, and the partnership to develop the player relationships that are key. Whether it is through marketing and/or better product insights, it is all about aiming to build stronger lottery player relationships across the channels. I think the starting point for lotteries is to consider their vision. Who do we want to be? What do we want to do? And then success is tied to working with the right partners to enable that to happen.
AM: I would add that I think a data collaboration with a vendor partner should result in lotteries benefiting from a more sophisticated, player-centric approach, from which players will gain an improved gaming experience. That’s why lotteries are engaging with a data platform. Using PDP is your strategic move to harness the full potential of data to inform decision-making and innovation. In addition to all the features, domain-level knowledge is crucial. It is important to note that something we have found is lottery data is not necessarily like any other data. It is not like retail data, for instance. The way people purchase in retail, they may research things and then they may go back and forth to a website. These patterns can be quite different than lottery because most lottery purchases happen around draws. When people come online to play their lottery numbers, they may play other games at the same time. Those patterns in the data and algorithms work that way; machine learning algorithms work from the data. The ability to facilitate all the features, centralized access, integration of diverse sources, real-time processing of data, and advanced analytics, are all important parts of a platform that a vendor can offer a lottery. Using the platform should result in something more sophisticated than they had previously.
SBC: What is the balance between AI and human involvement with the PDP platform? How much of this is automated, and how much is your team involved?
AM: There is a balance between being data-driven and data-informed. I think most organizations would want to be informed. The ability to present data to people so that they make their own decisions is critical. And we have that balance. We have those systems; we can show that and leave decisions in the hands of marketing or in the hands of lottery representatives. But we also have models that are more prescriptive in terms of providing a recommendation engine and churn prevention as well. Machines can get to a stage where they can instantly understand who is in front of them and provide those offers. But this should be rolled out at a lottery’s pace and must work with how comfortable lotteries are with these enhancements. We offer solutions for every different preference.
SBC: What is the most surprising player behavior insight operators have seen from PDP?
AM: It is a lightbulb moment, the way recommendation engines can work. The way the customer lifetime value model works of how we can predict things, some of the predictive analytics that we use that we have been able to show with remarkably high accuracy, such as right in the 90% range, how a game will perform from launch to the next 12, 18, 24 weeks to up and down trends.
KP: PDP can collect data on small events across the player journey and cumulate them into prediction models. These models give us a new level of actionable marketing information, like potential product cross/up-sell opportunities, player’s likelihood to redeposit or their predicted retention rate.
