On Perennial Sellers
push to talk #33 // why do some games just keep selling, years after release?
Editor’s Note:
Lately I’ve been alternating between essays and interviews + news, but for the next few issues expect a more experimental mix of formats and content types.
I’ll still be in your inbox every Friday.
Today’s post is an observational riff on the surprising marketing data you can extract from Steam by paying attention to certain under-appreciated metrics.
More on that below.
On Perennial Sellers
Every few weeks some goofy goober farts out a hit tweet by posting a screenshot of a game’s peak CCU chart alongside a snarky comment about “the fall” of a once-great game.1
The resulting discourse is always annoying and often flat-out wrong, as CCU is a limited way to measure success, particularly for premium titles or games in Early Access.
But in some sense, the industry’s obsession with CCU feels inevitable.
Valve doesn’t publish sales or revenue figures for games on Steam,2 and companies aren’t incentivized to share sales data unless they’ve hit a huge milestone. So CCU charts are the next-best thing we’ve got. Right?
Maybe not. Recently, I’ve become obsessed with a different kind of chart. They look like this:
Look at those beautiful blue bars. I love these things. This is reviews-per-week data for the 2023 remake of Resident Evil 4. Every bar on the chart represents total new reviews left by players over a 7-day period.
They’re wee little bars, thanks to the big blowout week RE4 had in late November 2023, but don’t let that fool you. The lowest weeks after September all brought in north of 400 new reviews. And the average week in the last six months has seen more like 600.
This might not sound impressive, until you realize that each review likely represents more than 30 sales.3
So let’s just do the math.
600 reviews per week implies (600 x 30) copies sold. That’s 18,000 sales per week, at somewhere between $29.99 and $39.99 each, depending on whether Capcom has a discount running. On the low end, then, you’re looking at $540,000 gross revenue every week, week after week after week after week.
This for a game that has already grossed over $100 million on Steam alone.4
We’re still in the early days for Resident Evil 4. It only came out last March! But I have a sense, based on the consistency of those reviews, that it’s in a class of games I’ve started calling perennial sellers.
A quick definition:
Perennial sellers are premium games (not free-to-play, though they may have live-service elements) with sales curves that seem to never go flat. Many sell thousands or even tens of thousands of copies every week, for years on end.5
This is not a phenomenon unique to Steam. Mario Kart 8 Deluxe on the Nintendo Switch is an infamous perennial seller for Nintendo. Every time Nintendo comes out with a public financials statement they sort of sheepishly admit they’ve somehow sold another 2 million copies of MK8. The game’s long lifespan was so unexpected that Nintendo didn’t release the first paid DLC for the game until seven years after it first appeared on the Wii U.
But that’s not too common in the world of console games. Something really unusual is happening on Steam, where many, many games are achieving perennial seller status.
Let’s look at some more examples. But before we do I wanna share three quick caveats about the data:
Something strange you’ll notice about these review history charts is that almost every game starts getting WAY more reviews starting in Nov 2019. That’s because Valve shipped an update to Steam around that time that increased how often players are asked to review your game. Average sales-per-review accordingly dropped around that time.
Many games get huge review spikes near the end of the year that are partly caused by holiday sales but also partly caused by the opening of player nominations for The Steam Awards. In order to nominate a game for an award, players have to submit a review for the game. So popular games tend to see a review spike around here that doesn’t correlate with more sales.
Technically when players update or change a review that also shows up on this chart, but that’s extremely rare unless you’re getting review bombed.
Got it? Okay. Back to perennial sellers.
First, the extreme high end of what’s possible:
This is Stardew Valley. Because the range is so wide, each bar on this chart actually covers an entire month. Since November 2020, there’ve only been two months when Stardew dipped slightly below 5,000 reviews per month. The median (I’m literally eyeballing for this) is more like 7,000, with many months bringing in 10k or even >15k reviews.
7k reviews a month implies a lowball estimate of 210,000 copies sold every month for nearly five years. This is for an 8-year-old, $15 game that is made by one guy.
Sure, he often discounts it to $8 or $9. But that’s still >$1.6–$1.9 million gross revenue every month for a game that came out when Obama was still president.
And here’s my favorite part: try to identify a curve on the chart. Where’s should the trend line, go? It just sells and sells, sometimes more (like with the recent 1.6 update, which caused the big sales lift this year) and sometimes slightly less.
Actually, we don’t have to guess about the slope. We can just look at it:
Stardew Valley is the definition of a perennial seller.
Next, let’s look at some well-known but not obviously huge games that are posting way crazier numbers than you might expect:
Damn, Deep Rock Galactic is kinda cookin’ right? And they’ve been on a particularly hot streak since 2023.
One thing that makes the Deep Rock story funny is their discounting strategy. The price history chart on SteamDB says it all, really. They just drop from $29.99 to $9.89 about as often as Steam will let them.
Here’s another good one. The multiplayer co-op island survival game Raft:
That’s a lot of castaways. Business has slowed down a smidge for them in the last couple of years but they’re still usually selling between 60k–100k copies a month at $13.39 apiece, sometimes $19.99. This game’s lifetime gross is over $150 million6 and I rarely hear other devs talk about it.
But enough talking about the monster hit games. The fact is, the vast majority of games on Steam are small titles developed by teams with fewer than 20 people.
And many of these—games that you’ve probably never heard of—are also inexplicably becoming perennial sellers.
Sleeper Hits
Exhibit A: Timberborn, a post-apocalyptic city-building game about beavers.
I gotta post the whole Steam description here because it’s just so good:
Humans are long gone. In a world struck by droughts and toxic waste, will your lumberpunk beavers do any better? A city-building game featuring ingenious animals, vertical architecture, water physics, and terraforming. Contains high amounts of wood.
I can’t believe they just casually slip in “lumberpunk” like that. Incredible.
And players seem to be digging it too:
This is a game that’s still in early access, which for the past two-and-a-half years has been steadily pulling in ~100 reviews (3,000 sales) per week, with no sign at all of slowing down. They never sell the game for less than $19.99, so it’s grossing at least $60k every week, apparently indefinitely.
For a AAA studio, that number looks like nothing. It’s a little over $3.1 million per year. But Timberborn’s developer, Mechanistry, is a small Polish studio with around 13 employees,7 so they’re doing great business with this thing—probably close to $250,000 gross per employee. In Early Access! In Poland!
Given the scale of the operation, this game is a monster hit. And it absolutely looks like a perennial seller in the making.
One more example of a sleeper hit, and then we’ll try to wrap this post up:
Pseudoregalia is an N64-style 3D platformer / metroidvania that absolutely rules.
It’s made by one person, who originally released a version of the game for free on itch.io after making it during a game jam. And it seems to now be selling a few thousand copies per week, every week, at around $5 per pop.
It’s only a year old, so it’s too early to call it a perennial seller, but I have a feeling it’s going to keep selling. It just has that vibe.
I love this example because it shows how the perennial seller phenomenon can operate at different scales, even for games with no live-service features. The pattern holds: The sales just kind of keep coming in on their own, regardless of what the dev does.
To explain this, you have to fall back on sort of hand-wavey ideas. Is it word of mouth? Sure. Is the Steam recommendation algorithm probably a huge factor in the ability of these games to continuously reach new people? Almost certainly.
But it begs the question: what makes these games different?
So What Makes a Perennial Seller?
If I look at the examples included here and other perennially selling games I’ve come across, I think they tend to have at least a few of the following qualities:
Extremely strong word-of-mouth – I’d define word-of-mouth growth for games as the thing that happens when extremely passionate fans of your game tell other people about it repeatedly and relentlessly, over a long time period. It requires evangelists, not mere fans.
Overwhelmingly positive review averages – I mean from players, of course, not critics. Of the games listed here, Raft has the lowest review average, with 93% positive on Steam. The others are between 95% and 98%. A score like this implies that a game has over-delivered at an near unheard-of level for its target audience.
A reputation as the best in its category – This sounds similar to the point above, but it has some specific connotations not guaranteed by high review scores. It’s not just that Stardew Valley is a great Harvest Moon style farming/lifestyle sim. It is undeniably the best one—even better than all the actual Harvest Moon games. The RE4 remake is an astonishingly great remake of a game long-considered the greatest horror game of all time.8 Pseudoregalia is probably the best N64-style Metroidvania.
This has a deep connection to the unnameable property that enables word-of-mouth growth for games. People play one metroidvania game and really like it, then they go online or to their friends to ask “what are the others I should play?” The best-in-category games like Pseudoregalia thrive in these moments.An experience you can’t get anywhere else – Deep Rock is this. (You’re a dwarf mining asteroids.) Timberborn is this. (Build a beaver empire.) Euro Truck Simulator 2 is this, if you are not European and/or don’t own a truck.
The most obvious quality I’m leaving out from this list would be many free content updates over time, which applies to many but not all perennial sellers.
On the one hand, it’s obvious that content updates can help sales over the long haul. On the other, for some games it’s kind of hard to disentangle the cause-effect relationship. Is Stardew Valley still getting updates because that would help it sell more, or is it simply continuing to sell at such a ridiculous rate that the dev feels compelled to keep updating it forever? Does this distinction matter? (No, lol.)
You’ll also note I left out anything about marketing. That’s because, as best I can tell, marketing doesn’t seem to matter much past these games’ initial launch. As a marketing director for video games you would think—to paraphrase Upton Sinclair—that my salary would prevent me from understanding this, but it’s just plainly true. Marketing can make a good game sell well, and a great game sell like gangbusters. But a game with the rare qualities of a perennial seller is going to sell itself even if the marketing is nonexistent.
There are almost certainly others shared qualities of these perennially selling games that I’m missing. I’d love to hear from you in the comments on email replies if you’ve thought of any.
That’s it for this week. I have a beaver empire to attend to.
I’ll see you next Friday.
My best guess for why Valve doesn’t publish sales or revenue figures is simply that devs would not like it very much. (Then again, they don’t love having their CCU charts made public either.)
I could write a separate article about the history of different approaches to analyzing sales metrics on Steam, and maybe I will at some point, because it’s actually pretty interesting and funny. But the plain fact is that “30 sales per review” is a good enough finger-in-the-wind estimate if you’re looking at a game’s reviews data and want to get a sense of how well it did. (It’s actually conservative as a guess: >40 sales per review is more common.)
If you want to get educated on this topic, the source of truth for people-in-the-know is this Gamalytic article from 2023 that broke down sales-per-reviews ratios across an wild number of variables. It examines why free-to-play games often go north of 100 “sales” per review. Or why very well-loved and niche games have potentially very low ratios (ratios as low as 15 have been reported).
So anyway, we’re just going with 30 here for simplicity’s sake. Please forgive me.
GameDiscoverCo’s estimate is $105,634,000. They lean a little conservative, which I like, so I usually refer to them as my source on Steam gross revenue estimates.
The “perennial sellers” phenomenon has been noticed by marketing people working in other creative fields. Its application to books was analyzed in an excellent book by Ryan Holiday called, fittingly Perennial Sellers, who I’m stealing the term from. I don’t love all of Holiday’s stuff (miss me with that stoicism ish) but I whole-heartedly recommend this book to anyone interested in marketing creative works.
$153,461,000 per GameDiscoverCo.
Mechanistry reported 13 employees in a press release last September.
Silent Hill 2 fans, let’s chat in two months after its remake comes out.
Love this! Definitely going to be looking at the reviews per week in my next case studies 🤙
Huh this makes me wish i was into "big data/data analysis". I can imagine doing some interesting stuff with just the public review info over time (and if positive) to determine all sorts of interesting stuff.