MICo CPO: Why AI Delivers Limited Returns for Operators

MICo develops machine learning solutions for iGaming operators across multiple markets. Its Player Intelligence product estimates each player's value and risk of churn, then recommends the next best action for every player. There is also a recommendation system for the game lobby. The company works with casino operators, platforms and white label providers.

CPO Aleksandr Kostiukov says MICo's products are not built on large language models or AI agents. Instead, the company trains its own models on the behavioral data of a specific operator. The models predict deposit and bet amounts. Based on these predictions, the goal is to raise metrics such as GGR, NGR and LTV through personalization. Kostiukov notes that traffic has become more expensive and winning players has become harder. For this reason, he says, the products work with the existing player base and increase LTV without additional marketing spend.

Kostiukov says the core benchmark for each model is a transparent financial increase in the operator's own figures compared with a control group. He adds that results which cannot be mathematically verified in an A/B test are not counted as successes. The company's system reportedly works alongside the operator's CRM, with the models passing player scores into the environment teams are already used to. On the data side, he says personal data, contact details and payment information are not required. Anonymous identifiers, location, deposit, betting and session data are said to be sufficient. It is also noted that data is not merged between clients.

According to NEXT, a recent study by UNLV and KPMG found that only one in five gaming companies reported meaningful returns from AI.