Last October, Steam’s algorithm briefly went haywire, filling users’ “More Like This” recommendation sections with popular games while negatively affecting under-the-radar indies who needed the exposure. At the time, Valve chalked this up to “a bug” which it said it corrected, but smaller developers complain that Steam’s recommendations have continued to let them down ever since. Now Valve says it’s changed recommendation sections to de-prioritise bigger games and focus more on personalised selections for individual users.
Valve began its announcement post by admitting that recommendations had some issues. “Previously, when customers would look for games by browsing the recommendation feed at the bottom of the homepage or the ‘More Like This’ sections, they weren’t seeing as many different games as we would’ve liked,” the company wrote. “Furthermore, we were receiving lots of feedback that ‘Recommended for You’ felt too biased towards only the most popular games and didn’t feel very personalised.”
So Valve went on a bug hunt.
“We found some bugs, such as the ‘Similar by Tags’ section of the Recommendation Feed, which had a bug that top-rated games (a category that doesn’t change very often) were driving too much of what players saw,” Valve wrote. “We changed that. We also found that in some places our timescale used to calculate popularity was too narrow, resulting in unpredictable visibility for some games. So we expanded the time period we use in those calculations.”
These changes should now be available to all Steam users. Valve tested them on a small segment of users prior to launch and found that the fixes resulted in users clicking on recommended games more often and visiting the pages of a greater number of individual games.
“In these changes, ‘Recommended for You’ became less biased towards popular games, and showed games that are more relevant to individual customers,” Valve wrote of the algorithm test. “As it turned out, customers in the experiment group were more likely to click on the games shown in the recommendations section, at a rate almost 15 per cent higher than the control group. The increased personalisation means there is an even greater variety of games being shown in this section, and customer impressions are more evenly distributed among them.”
There was also a 75 per cent increase in the number of unique games visited, and a 48 per cent increase in average visits per game compared to a control group with the prior popular-game-loving, Mountain-Dew-chugging version of the algorithm.
For years now, Steam has had trouble surfacing smaller games to users who might be interested in them. This has frustrated indie developers, many of whom now see potential success on the platform as a crapshoot even if they generate a fair amount of buzz around their game pre-release.
Recommendations have been an especially big thorn in developers’ sides. Between this algorithmic change and experimental new tools like an AI that scans your playtime and helps you figure out what to buy next, it seems like Valve is finally making a concerted effort to minimise the problem.
Valve believes these early results signal a change for the better on Steam.
“We’re encouraged by these results and have now rolled them out to everyone,” the company wrote. “We continue to make changes and run experiments like this in order to improve Steam’s existing features, while we also explore entirely different ways for customers to find games they love.”