Steam with the help of machine learning began to suggest suitable games to users

The service is based on the number of hours that a particular user has spent in certain games. Then he finds users with a similar gaming experience. In the output, the service suggests to a specific user those projects that he does not have, but which gamers like him have played for many hours.

The output can be customized according to the degree of popularity of games, their age, as well as highlight or exclude categories (for example, make sure that free or indie games are not indicated in the selection of suitable projects).

The main problem of the tool is that it is difficult to find it. Until it is placed on the main page, many are unlikely to use it. On the other hand, while he was part of the Steam Lab test program, thanks to him, more than 10 thousand different games were bought in the store (meaning not the number of copies, but the number of titles).

Plus, Valve is starting to use this system in other sections of Steam. For example, when viewing a particular game, one or another user will be able to see the phrase “Players with similar preferences like this game” if the title suits him.

As a result , the authors of the system identified the following:

• the collected data showed a strong and stable advantage of the “Interactive Assistant” over other functions on Steam;
• the positive effect of its introduction is long-term, in other words, the function has a long-term perspective;
• with its help, the lion’s share of the entire Steam catalog is recommended, not just recognized hits.

Also on the topic:

• A news feed on products of interest to players will appear on SteamValve has changed the game search system on Steam
• Game developers for Steam have received a tool for re-engaging users
• Is there any news?