Artificial intelligence (AI) is a top priority across boardrooms in every industry, and for good reason: PWC estimates that AI could contribute up to $15.7 trillion to the global economy by 2030.

The gaming industry and casinos are no different. Beyond competing with other casinos for a player’s entertainment budget, they are now competing with an ever-growing list of entertainment options from online gaming and streaming to traditional movies and television.  Casinos have troves of data on their customers that can inform their AI efforts. This data used to be an advantage over alternative forms of entertainment, but the other industries are catching up. It is a race to leverage this data and use it to drive revenue.

This can be seen in the way Netflix uses data on how users interact with its service. By tracking all the information about what shows its users have watched, when they paused or stopped watching, and combining that with information about the type of show, Netflix can better decide what shows to invest in. Likewise, casinos have similar information about their patrons. In order to compete, casinos need to leverage this data just as Netflix does to drive additional patron engagement, loyalty and play.

Netflix and companies like it employ massive numbers of expensive, highly-educated data scientists to produce these results. Fortunately, with the rise of automated machine learning platforms, casinos now can begin leveraging AI by using their existing staff. By developing their AI capabilities intelligently, they no longer need to invest in huge teams of data scientists.

Casinos should be cautious, however. Numerous reports indicate how often AI efforts go awry.  A report by IDC earlier this summer showed that organizations were seeing a failure rate of up to 50 percent in their AI projects. In 2018, Gartner predicted that up to 85 percent of AI projects will not deliver for CIOs. But casinos should not use this as an excuse to avoid developing AI solutions. There are ways to mitigate this risk and compete with the entertainment options already deploying AI. Casinos just need to follow a few simple steps.

First, casinos should start small and seek quick wins. They need to focus on the areas that already have plenty of data that is already being used to drive business decisions. The obvious answer is slot machine loyalty marketing. Casinos are tracking player behavior and using it to make marketing offers to players today. With a reasonable amount of work and help from automated machine learning solutions, database marketing staff should be able to develop AI solutions that can identify the right offer (free play or property discount) at the right time to drive the maximum future play for every member of their loyalty club.

Casinos should be careful not to try to immediately turn this into a major system that will predict in real time what offer, if any, to make to each player at any time. Instead, they should focus on what they are already doing. If the casino’s database marketing team is producing monthly mailings, use an AI system to improve that first. In the first iteration, simply use AI to determine what offers to send to each player. The advantage of this approach is that most casinos are tracking the responses to their marketing offers and can readily evaluate the performance of the AI and modify it quickly to tune its performance.

One additional advantage of this approach is that it will keep the AI functioning in the language of business. In many cases where an AI system has failed, the system was developed by data scientists who failed to understand the business, either due to hubris or lack of time. As a result, they developed systems that suggested solutions that were either impossible to implement or were explained in the language of data science and, therefore, not understandable by those on the business side. Allowing existing staff with a knowledge of the business and what’s possible within marketing to develop the AI systems keeps AI projects scoped to what will drive actual business value and avoids solutions that are impractical.

Finally, focusing on a specific project avoids two other common points of failure in AI projects. First, this allows the AI project to actually be completed in a timely manner. Many AI solutions fail after spending years and millions of dollars. Second, this project has a clear method of turning the output of the AI into business decisions. Too often, AI projects have vague outputs where the purpose is not clear.

With the proliferation of both entertainment options and the use of AI to drive business, casinos need to begin thinking about how they are going to incorporate AI into their entire decision process. Developing an AI roadmap is daunting, but by starting small, focusing on data-driven areas they already work in, and leveraging automated machine learning tools, casinos can extract value out of their data and thrive in the new entertainment landscape.