Steam retires manual curation: sales and events will be handed over to the algorithm

Valve is preparing one of those changes that seem small but end up changing how we discover games. According to Gamereactor Spain, the company plans to replace Steam's Sales and Events spaces, which are currently managed by hand, with algorithm-generated personalized recommendations. The change is expected to arrive in early 2027, although Valve has not yet confirmed a specific date.

Why does it matter? Because that section is one of the main storefronts of the biggest PC store in the world. Until now, what you saw there depended on promotions scheduled by Steam partners or on what Valve's team picked directly, always within a limited number of slots. With the new approach, each user will see a different selection tailored to their tastes. For players, it's a new way to find deals; for studios, it's a new lottery for visibility.

What you need to know

• The Sales and Events section will move from manual curation to personalized recommendations.
• The rollout is expected in early 2027, with no official date confirmed.
• In testing, the algorithmic system showed users 10 times more games in that section in a single day.
• Valve reports significant increases in store page visits, wishlist additions and items added to cart.
• The Featured & Recommended tab isn't going away and will keep mixing different types of picks.

What changes and who it affects

The change is very specific: it only affects the Sales and Events section. Steam's main front page isn't suddenly turning into an endless social-media-style feed, at least based on what has been shared so far. But the logic is clear: less of a fixed storefront for everyone and more of a personalized showcase for each player.

Valve's numbers are worth reading carefully. The system showing ten times more games doesn't mean ten times more sales, or that total traffic will grow tenfold. The report itself clarifies that the results reflect overall activity during the experiment, not a guarantee of equivalent growth in sales or visits. It's an improvement in exposure and engagement, not a revenue promise.

Studios at the center

Developers and publishers will be the most affected. Today, landing one of those limited slots can make or break an indie's weekend. With the algorithm, the exposure will be spread more widely, but it will also be less predictable: your game will be shown to whoever the system thinks is interested, not to everyone at once.

Our take

Here in the Underworld Lordz newsroom, we see it with a mix of excitement and skepticism. The upside is obvious: more games getting visibility and less reliance on a handful of slots that usually end up going to the same old names. If the algorithm works well, more than one competitive indie with a small community could finally reach players who didn't even know it existed.

The downside is easy to see coming, too. Recommendation algorithms tend to lock you into what you already play. If your history is full of tactical shooters, you'll probably see more tactical shooters, and that makes discovery poorer. Content creators face another problem: if every user sees a different store, the shared storefront that sparked videos and streams like "the deals you can't miss" disappears. The conversation around sales becomes fragmented.

And there's a deeper criticism: human curation, for all its flaws, had judgment and visible people accountable for it. An algorithm optimized for visits, wishlists and carts rewards what hooks you, not necessarily what's worth your time. We're asking Valve for transparency about how the system makes its decisions, and for tools that let players fine-tune their recommendations.

What do you think?

It's more than a year until this arrives, if the schedule holds, so there's time for Valve to listen to the community. We've already made up our minds, but we want to hear yours. Would you rather have a sales section tailored to you by an algorithm, or the classic hand-picked selection? Let us know in the team chat or on our Discord.