Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Creative company emerges today as a new type of market intelligence company. He uses AI to analyze feelings on 1.5 million conversations on the best game publishers and their titles.
This means that he uses AI to understand what players think of 17 of the best game publishers – with the ideas produced by Machine Learning. The Creativ AI was able to analyze more than 1.5 million online conversations on Reddit, YouTube, Discord and press items over six months. It took about 10 days to do it. The company has known about 9,300 sources of news and feeling on game publishers. Then, he did his analysis for the study, which covered the period from November 1, 2024 to the end of April 2025.
I spoke with the CEO of Creativ, Wes Morton, the CTO Joe Lai and Coio (chief of the operation and information manager) Vibhu Bhan on the results. Here is some of the information from the exclusive analysis.
“We call ourselves an intelligent marketing company.” The study is literally an ingestion of a million conversations and a half consumers on these different publishers. “
In the extraction of “the analysis of feelings”, the objective is to know what players think of game companies according to what these players say on social networks. The large language model (LLM) of the AI is formed to detect sarcasm, the argot specific to the game and more nuances, said Bhan.
“Real innovation here is a better understanding of the context and the slang, so the analysis of feelings is much more contextual and not just a score,” said Bhan. “If you look at the analysis of traditional feelings, it examines the existence of certain words. But language is complex.”
The analysis of feelings has appeared in recent years as a means of understanding the zeitgeist around a game or a business. But often, the analysis suffered because the analysis used did not really understand players or their comments on subjects. Now, with LLMS, Morton said that automatic learning includes complex nuances and does a better job on more data than it can ingest.
In an example, Creativ noted that fans were not satisfied when actor Henry Cavill was dismissed from Geralt’s main role in the Witcher television show on Netflix. Basically, Netflix should not have drawn Cavill, as this led to a global negative impact on the Witcher franchise. It turns out that the show influenced the general feeling, rather than the video game series.
The company ingests the data and then offers feeling scores on game publishers to see what they have done to help or hurt their brand lately in the conversation with the players. Ancient reports could understand how often a set of words (such as a game or business name) has been used. But he often did not have the ability to understand the full context around a discussion on games, then summarize it correctly. But LLMs are better to understand the context around a large amount of data.
“The context becomes much more important because it allows you to understand the direction of feeling because there could be some subjects in the sentence. And the second thing is this switch that we do because of the sarcasm, which is perceived as a false positive when it is designed as a negative reaction,” said Lai.
LLMs have a better ability to understand the context of the language, said Lai.
“And the beauty of the LLM is that we are able to collect and train our models on this game data,” said Lai. “We are able to train the models to be able to detect the information line that appears for each of these games, as well as if they are used positively or negatively.”
The biggest conversation subjects
One thing that the LLMs took care of is that players had strong opinions on exclusives, and if a platform owner should keep his best exclusive game or take this game on other platforms in order to generate more sales. Fans who have invested their money in a particular console did not like it.
The biggest conversation subjects included the monetization of the game, the franchises, the game platforms, the exclusives and the consolidation and the corporatization of the industry. In monetization, the players rewarded the communications open to the rules and the studios which avoid the models of monetization which affect the gameplay and the mechanisms. It was the widest trend in the data set, consumers perceiving Activision Blizzard, Ubisoft, EA, Amazon, Netease, Evolution Gaming and Roblox as delinquents particularly bad in bad monetization practices.
In addition, the LLMS capture the conversations that occur naturally. On the other hand, a study puts the player on alert that they are questioned for their opinions. This player could think about the opportunity to respond honestly or not, depending on what they think that the study researcher wants to hear.
How companies behaved
Netflix did not have a lot of history as a game publisher, and its mobile games have not yet been huge success. This helps to explain why he obtained a negative score of players. Part of the feeling occurs around a game, just like an NBA match, but a large part occurs outside the game on social networks.
Morton said that games get a great increase in Hollywood awareness, because movies based on games like a Minecraft film and the TV show The Last of US get high notes and reach more people who do not know games.
“The cool part of this technology is that you can specifically explore what makes people happy, sad,” said Morton.
Activision Blizzard had a lot of chatter on World of Warcraft. But many players were not fans of how the company managed the Overwatch transition to Overwatch 2. Ubisoft also released the worst score of all game publishers, but we didn’t know why. He had a lot of discussions on the characters of Assassin’s Creed: Shadows. But this game received positive criticism unlike previous games like Star Wars: Outlaws and Skull & Bones.
For this study, the company has not focused on any particular game. But that could do this in the future.
With LLM, the study can be carried out in 10 days, compared to weeks for other methods. Morton said LLM can only absorb and ingest and process data faster, but can analyze much more data and much faster. Over time, the analysis can become much more granular, emphasizing the characters or other details of a given game. Such an analysis could give a team a chance to rotate another character if she has a negative score.
