Perthro

Stories··7 min read

The best game recommendations come from people

Store pages know what you bought. A friend knows why you liked it.

Think about the games that really mattered to you. Not the ones you played because they were everywhere, but the ones that stuck. There's a good chance at least one of them reached you because a specific person told you to play it.

That isn't just nostalgia. In Google's 2022 survey of more than 14,000 PC and console players, recommendations from friends and family were the second most common way people discovered games, at 38%, just behind YouTube at 40%. Store dashboards trailed far behind. And some of the biggest hits of the last few years, from Among Us to this year's Meccha Chameleon, grew almost entirely through players telling other players. Here's why recommendations from people work so well, and how to give and keep better ones.

How players actually find new games

Surveys disagree on the exact numbers, and they all let people pick more than one answer. But friends consistently show up near the top.

SurveyWho was askedFriends / word of mouthVideo (YouTube)Stores and dashboards
Google for Games, 202214,356 PC and console players, global38%40%16–17%
Big Games Machine, 20241,009 US console and PC players34%64%n/a
YouGov, 2023US adults who play an hour a week or more17%n/a4% (in-store)

The YouGov figure is a useful reminder that word of mouth isn't the only route. In that survey, social media (22%) and online reviews (14%) ranked alongside it in the US. But when the question is which channels players trust and act on, people they know keep coming out ahead. In Nielsen's global trust in advertising survey, 83% of people said they completely or somewhat trust recommendations from people they know, more than any other source, including branded websites (70%) and online reviews from strangers (66%).

Two friends on a sofa playing a game together
Two friends on a sofa playing a game together

What a store's algorithm knows about you

Store recommendations are built from your behaviour. Valve's own documentation says Steam's main store capsule "is different for every customer," based on "games they own, games they play, games their friends play, top sellers, etc." Its Interactive Recommender, launched through Steam Labs in 2019, is a machine-learning model that looks at how much you've played each game in your library and was trained on "many millions of Steam users and many billions of play sessions."

That's powerful, and genuinely useful. But it's built to find games similar to ones you already play. If you've played three roguelikes, you'll be shown a fourth. It can't easily know that you love tactics games and slow detective games for the same reason: both make you think before you act.

What a friend knows about you

A friend recommending a game is matching it to you, not to your play history. They know you hate time limits, that you'll forgive a rough first hour for a good story, or that you've had a stressful month and want something gentle.

They also turn a game's reputation into something human. "It's a 90 on review sites" is a number. "I stayed up until two in the morning on a work night" is a story, and stories are what make us actually press start.

And a personal recommendation doesn't end when you buy the game. When you finish it, you have something to talk about: favourite moments, the ending, what you'd play next. The next recommendation you give them is shaped by that conversation.

The biggest word-of-mouth hits

Some of the most successful games of the last decade had little or no marketing behind them. Players did the work.

What these games have in common is that they're best with other people. Many of the biggest word-of-mouth hits are games you can't really enjoy alone, so recommending them is also an invitation to play together.

Why good recommendations get lost

The problem is that personal recommendations are fragile. They happen in passing: in a group chat, at a party, in a comment under a video. By the time you're looking for something new to play, you can't remember who suggested what, or why.

That's a shame, because the why is the valuable part. "Sam said this had the best ending they've ever seen" is worth more than the game's name on its own. Without the context, a recommendation becomes another item on a wishlist, and wishlists are where games go to wait forever.

How to give a better game recommendation

How to keep the recommendations you get

When someone recommends a game, write it down with their name and what they said. It takes ten seconds and turns a passing comment into something you can come back to. And when you finally play it, tell them. Most people love hearing that a recommendation worked.

This is a big part of why we built Perthro around friends and not just a list of games. Seeing what the people you trust are playing, and what they actually thought of it, is still the best way we know to find your next favourite game. We wrote more about why game communities matter if you want the longer version.

Sources

Keep reading.

Every post →
Stories4 min read

Should you replay GTA V before GTA 6?

Stories7 min read

What to do after you finish a game you loved

Stories7 min read

Why we replay the games we love