Some Motivation:
Being able to predict sales of a game is a really useful tool. It tells us how much to invest in our own projects, it can give insight into what aspects are worth investing in or emphasizing in our marketing, and it allows us to give better feedback towards others.
Obviously game markets are too complicated to do this with much accuracy, but we can practice to be better. There's a couple ways to do this:
When we experience surprise at a result it's often a good idea to ask ourselves what we didn't consider. Surprise at a result can be a good sign that there's something we don't know, that our mental model is off, and so noticing surprise is an opportunity to reflect on the difference between what we expected and what we observed.
If we predict beforehand what our expectations are then we can then accurately track how far off we were, and knowing that gives us insight into how much we need to adjust our mental models. Quite often it's simply a case of not knowing some factor, but other times we'll be surprised even with enough information, in which case there is something we don't understand.
The Game:
The idea is that when a game is due to be released, be it a game of our own, or a game that is relevant to us (because maybe it is similar to a game we'd like to make or are making), we should suggest it here and state our sales estimate.
Anyone from this community should suggest a due-to-be-released game, and other people here can add their estimates. We can also give brief explanations of why the estimate.
We'd then track the sales on SteamSpy ( http://steamspy.com ) and see how far off we are. When we're far off we'll know our inaccuracies for certain and we can try figure out as a community what it is we don't know (assuming anyone has any idea what they or anyone else doesn't know).
The sales estimate should include a timeframe, for example "sales in the first 3 months". Digital sales do continue on for a long time, but after a certain point deep discounts start happening and games go into bundles which make the sales data much less meaningful. And in any case we don't need to guess lifetime sales in order to judge our mental models (and lifetime sales will of course take forever to evaluate against).
This isn't about wishing poor sales or excellent sales on each other and other games. A low estimate doesn't mean we want a game to fail, it just means we expect it to sell that amount. If we're guessing too low it just means we don't know or don't understand something about the game or marketplace (and we can help each other correct our understandings and fill in our gaps of knowledge). The same goes for if we're guessing high.
Neither is this a competition. I certainly just want to learn to know how well a game is going to sell, and, more importantly, why it is going to sell, in order to make games that more people want to play and give better advice to others.
Being able to predict sales of a game is a really useful tool. It tells us how much to invest in our own projects, it can give insight into what aspects are worth investing in or emphasizing in our marketing, and it allows us to give better feedback towards others.
Obviously game markets are too complicated to do this with much accuracy, but we can practice to be better. There's a couple ways to do this:
When we experience surprise at a result it's often a good idea to ask ourselves what we didn't consider. Surprise at a result can be a good sign that there's something we don't know, that our mental model is off, and so noticing surprise is an opportunity to reflect on the difference between what we expected and what we observed.
If we predict beforehand what our expectations are then we can then accurately track how far off we were, and knowing that gives us insight into how much we need to adjust our mental models. Quite often it's simply a case of not knowing some factor, but other times we'll be surprised even with enough information, in which case there is something we don't understand.
The Game:
The idea is that when a game is due to be released, be it a game of our own, or a game that is relevant to us (because maybe it is similar to a game we'd like to make or are making), we should suggest it here and state our sales estimate.
Anyone from this community should suggest a due-to-be-released game, and other people here can add their estimates. We can also give brief explanations of why the estimate.
We'd then track the sales on SteamSpy ( http://steamspy.com ) and see how far off we are. When we're far off we'll know our inaccuracies for certain and we can try figure out as a community what it is we don't know (assuming anyone has any idea what they or anyone else doesn't know).
The sales estimate should include a timeframe, for example "sales in the first 3 months". Digital sales do continue on for a long time, but after a certain point deep discounts start happening and games go into bundles which make the sales data much less meaningful. And in any case we don't need to guess lifetime sales in order to judge our mental models (and lifetime sales will of course take forever to evaluate against).
This isn't about wishing poor sales or excellent sales on each other and other games. A low estimate doesn't mean we want a game to fail, it just means we expect it to sell that amount. If we're guessing too low it just means we don't know or don't understand something about the game or marketplace (and we can help each other correct our understandings and fill in our gaps of knowledge). The same goes for if we're guessing high.
Neither is this a competition. I certainly just want to learn to know how well a game is going to sell, and, more importantly, why it is going to sell, in order to make games that more people want to play and give better advice to others.
Comments
Broforce is about to leave Early Access. I'd like to guess how many sales it will receive in the first two months after launch.
To date it has received about 357,594 sales on Steam according to SteamSpy: http://steamspy.com/app/274190
My prediction is that it will sell a further 115,000 copies on Steam in the first two months after launch.
This is mostly based on the Crypt of The Necrodancer figures. Crypt sold an extra third of its sales in the first month after its launch out of Early Access. I think Broforce will do a bit worse because the holiday season is a worse time to launch and Broforce doesn't have the strong replayability that Crypt does. I'm a little concerned that the final version will disappoint and push down sales. but I've never been able to predict the reception of my own games well.
The first two months after Broforce launch should not include any Christmas discounts either (if there are Christmas discounts they will be happening after the first two months).
Devolver on the other hand expect more than that (I believe). In the past they have been more optimistic than me about Broforce and they've also been more right.
I know there isn't much data on games launching out of Early Access, and these launches have had vastly varying circumstances. But if anyone else would like to weigh in and make some predictions I'd be interested.
Reasons:
-Game has stellar reviews on Steam, that's going to drive a lot of people to it now that it's out of early access and complete in their eyes.
-You have Devolver pushing PR, I suspect they think it'll do a lot more than our predictions.
-Another round of Youtube attention can't hurt, release is a big press milestone as well.
-Holiday periods aren't bad, especially if you can get into the top lists on Steam during them (which you will) I reckon you won't drop out of those lists very fast at all.
-If you're doing a launch discount... Well. Steam users like the green boxes.
Again cause of the youtube-ability, devolver and word-of-mouth of broforce.
It's a quite highly praised puzzle game, with a few similarities to World of Goo, but I'd argue, much less charm. It's been an IGF finalist for "Excellence in Design". It's also been at Indie Mega Booth etc, so there's been some good exposure to the audience that probably wants this game.
Mushroom 11 has been funded by Indie Fund, so there's a few pretty influential people backing it.
I really am not sure of how well this game is going to sell. I think it's got a lot of the elements that made World of Goo successful, and is being backed by the creators of World of Goo, but I don't think it has ALL the elements that caused World of Goo to be such a success, and the market is quite different now.
So I'd guess, in the first two months (before Christmas sales) it'll make 40,000 sales. That is two times what Lovers in A Dangerous Space Time is looking like it'll make, so I might be guessing a bit high.
I'd guess a 20% bump on it's current sales of 1,395,176 (according to Steam Spy http://steamspy.com/app/233450 ) during the first two months after launch, and I'm guessing lower than Broforce's bump because Prison Architect has done many deep discounts already https://steamdb.info/app/233450
I'm curious about this one because they are looking to get some of the audience of Don't Starve. And their visual style channels Edmund McMillan a little bit (that a bit wackier and more colourful). It seems like a thing that should work, though I don't think they're putting the level of polish in that Don't Starve has, and their marketing message is a bit loopy (but entertaining).
But they do have a following, possibly mostly on mobile. I really don't know what how much this is going to affect PC. If the game is perceived as a mobile game that could also hurt it on Steam (I think?)
Don't Starve did brilliantly, and the style of game is designed to build an audience over time. Guess follows -> I'm going to predict 20,000 sales in the first two months, but I'd expect a long tale.
I predict Broforce will sell 50k copies in the first two months after launch.
I predict Prison Architect will sell 200k copies in the two months after it launches, but I also think they will discount relatively heavily in this time, in which case they will sell more.
I know a lot of devs say it gets close, but it seems to be reporting a far lower number for Mike Bithell's Volume than I'd have expected. It's still sitting at around 5k.
It's probably just an aberration, though, for the most part devs have reported SteamSpy as being fairly accurate.
SteamCharts says Volume has had peak players in its lifetime of 100, and no higher than 17 for the last two months, which is inline with SteamSpy's figures. Obviously these figures will rise still, Volume has never been lower than a 33% discount, nor been in any bundles yet, as far as I can tell from SteamDB.
The game has a 88% userscore on Steam, which is very low for an indie game who's audience cares about game design and much lower than Thomas Was Alone (at 95%). This for me is a very important point of data, if there is more than double the amount of reasons to recommend against purchasing Volume than Thomas Was Alone then I'd expect an order of magnitude worse sales. Obviously we can't guess the user reviews before the time, but Volume always seemed less loveable to me (for a variety of reasons).
[ Edit ] Just saw the Steam Spy Venn diagram feature about Volume and Thomas was Alone... Volume has also apparently been searched for a ton since then (6th most searched for game on Steamspy), so I guess a lot of people are surprised.
@garethf Why would you have thought much higher for Volume?
An interesting counterpoint to Volume seems to be that TWA looks like it's sold at least 15K copies since Volume was released. If true, how does that impact the predictions of success for Volume?
Another interesting thing is that Keep Talking And Nobody Explodes has a really epic sales curve right now. Go look :)
It's also got a 99% positive rating on Steam, and an almost perfect Metacritic score. But it's novelty has to be a big factor I think.
Undertale is the other game that has recently come along and surprised me with its success. Also a super well loved game.
Because of this:
https://twitter.com/mikebithell/status/652040045629702144
I'd expect a higher number than 5k for something that outearned three quarters of the release year of TWA in 2 months. It might just be selling mostly on consoles, but it seems odd.
I don't have enough information to know whether it's accurate or an error, and if it is an error, whether it's a rare aberration or a common error we're just not aware of.
I've heard a number of devs confirm that SteamSpy has been pretty close for their games, but I don't know what percentage of the whole the affirmations are. Are there are lot where it's off the mark and the devs are just keeping silent for whatever reason? I just don't know.
So I'm cautious.
1 Mike's talking about earnings (which I'm going out on a bit of a limb and am assuming is the gross income of the game after platform share deduction)
2 TWA and Volume have different prices $10 for TWA and $20 for Volume. So Volume would only need to sell half as many units as TWA at full price to make an equivalent amount of money, which would explain a lower number.
---
An unrelated related fact is TWA didn't do THAT well at launch. A lot of it was incremental building of figure and such through multiple releases. These incremental releases had a second order effect that they lead to more publicity and thus more sales for the already available platforms. The notion of TWA being popular probably has more to do with it having been released multiple times over multiple years than anything else.
One other thing I'd like to talk about is that Mike is just mentioning earnings being more. Looking at the development of both games: Volume cost a lot more to make. I'm not sure if Mike's said anything about the game breaking even or turning a profit, but I think even with the given price difference it could potentially take longer for Volume to be profitable than TWA. Just a note that this last paragraph is thumbsuck and speculation, but I think it's worth mentioning just so we can have a better understanding of what we understand as success and how a single datapoint doesn't really give us a picture of what that is.
And because number and OMG indiepololypse because look at the numbers! Is all the rage these days lets look at some numbers for TWA, I'll talk about how Volume relates to this #TotallyAwesomeAndFactualKnowledgeOfAllThisWonderfulDataOMG
TWA launched 2012-06-30 on Desura
TWA launched 2012-11-12 on Steam
Which is about 4.5 months of 'launch' without being on the biggest platform in the market for PC distribution, and is also half of the 9 months of TWA that Volumes 2 is supposed to be equal to. Of course this operates on the supposition that TWA's launch that Bithell is referring to is the Desura one as opposed to the Steam one.
Ok now lets look at TWA as a 'success'
2014-04-25 it was reported that TWA shipped a million units. While this is totally cool, it's not exactly clear what that actually implies.
I'm just going to look at the game until this point.
So firstly TWA was in Humble Indie Bundle 8 (2013-05-28). The bundle sold 448,750 units and TWA was in the lowest tier so everyone who purchased the bundle got a copy. Looking at current HIB workings 65% of what you put into a bundle goes to the devs, and that seems to be an even split amoungst all tiers of games. The bundle had an average purchase price of $5.86 with 11 games in the bundle, and using the power of maths we can tell that TWA made $242,170 for 448,750 units or $0.54 a unit.
TWA was also in a weekly bundle (hosted by pewdiepie 2013-08-15) which sold 189,927 units (no tiers for the games) at an average of $2.87 per unit with 5 games in the bundle. Means that TWA made $70,862 from the bundle at a value of 0.37 per unit.
Now TWA launched PS3 and Vita 2013-04-23
So between the launch on Desura in mid 2012 and crossed a million units in April 2014 and of that million 638,677 were in bundles at that point.
---
Now there is one important fact that we're missing here: Volume dual launched PC and PS4 at the same time. At it's got some pretty decent support from Sony for the launch. So the sales that Bithell is talking about, I assume, include the PS4 sales which, to the best of my knowledge, we have no idea about.
tl;dr we can look at all the numbers we want but ultimately we're dealing with a really fucking limited set of data when it comes to games. And inference is a pretty shitty way of working things out, but unfortunately it's the best tool we have to try and analysize these things, which is why the analysis drawn from OMG NUMBERS!!!11! (read indiepocolypse articles) kinda really miss the point, and guessing games aren't that useful.
On October 29 GalakZ is coming out. On PS4 it received 82 on Metacritic with a 7.2 out of 10 user score. I've actually played this game, and didn't find the Roguelike elements nearly as compelling as reviewers suggested they would be, nor the gameplay nearly as tactical as the systems were claimed to facilitate (I think that the marketing material caused a bit Placebo Effect affecting reviewer's perceptions)... But the game has pretty solid arcade combat, and looks very pretty, and has shown to have some savvy marketing behind it at every turn. I think a lot of fans of retro games are going to pick it up, there's several nostalgia hooks in there, but that the game won't have much traction outside of that.
I'm putting my guesses under spoilers (so other folks here can make their own independent guesses).
I'm guessing that in the first two months it will sell 25,000 copies on Steam (a bit more than double Lovers in a Dangerous Space Time) and will have a 94% positive rating on Steam.
We (at Free Lives) are interested in games where the player builds and destroys things with the things they build. So we're interested in how well Forts does on Steam (and figuring out why it succeeds or fails as much as it does).
My predictions (both for if it has online multiplayer and if it doesn't):
Based on what looks like a fair level of polish and execution, but borrows from mobile games for its aesthetic, and that the building system works a lot like World Of Goo. I think there is potential for this to be very enjoyable, but I think the audience appeal is limited by the army theme. I think the overlap between World of Goo players and army game players isn't large. Worms sort of hits a middle ground in terms of audience, but it doesn't look like Forts is going to achieve that kind of charm.
If the game is local only, I expect it to reach 14,000 sales in the first two months.
If the game has online (which I think is unlikely because of the physics) I'd expect 35,000 sales (Broforce did 60,000 sales in the first two months on Early Access).
anything in the 50,000 to 100,000 range is possible
Here's the trailer (which doesn't include sales figures so far thankfully):
I'd guess this game is going to sell 6000 copies in the first two months. It might actually be quite fun (as people who played it at the Indie Megabooth at PAX thought), but the theme is basically the worst parts of Indianna Jones 4, and the polish just isn't there. I think I might be overestimating in fact, even if they have some discounts in that time.
http://www.twitch.tv/ryan_clark/ at 1pm pacific time tomorrow.
Broforce sold 90,786 copies in that time. Nearly everyone over estimated, with only @raithza predicting lower at 50,000 copies.
Broforce was in an Autumn sale, which added about 30,000 copies. I didn't expect a Steam sale so soon after launch, so my estimation was about 2X more optimistic than it should have been.
Prison architect is sitting at 1,480,246 sales on Steam Spy. Which is about 85,000 copies sold since launch. I was about 3 times too optimistic about that.
What this tells me is that exiting out of Early Access doesn't mean much, but it's the discounts that drive sales. Prison Architect had done 75% discounts before launching, and it appears like it's just kept on selling at the same pace post launch (which is a good pace nevertheless).
Mushroom 11 has 3,309 sales on Steam Spy. I predicted 40,000 sales, which is obviously WAY too much.
I think to an extent, trying to predict the first two months of sales has thrown our estimates off a bit. Steam sales have a long tale, and we usually think of games as having year long cycles, and Mushroom 11 has still got all its discounts to go. However 10X overestimating is very inaccurate.
Once again, @raithza's estimate was more pessimistic, at 20,000, but still not nearly pessimistic enough.
I think some of these 2 month predictions might be more accurate for a year of sales. Maybe we should be predicting the first two months, and then also predicting a year of sales (which might help us calibrate a bit better?)
They're also very likely to be in a Christmas Steam Sale, which should get them 20,000 sales on its own. In the first two months I'd expect 80,000 sales. I don't know if that is pessimistic enough, but if they have done no discounts up until this point, and have around half the sales, I'd expect they should hit figures at least as high as the Broforce launch. Also it's a very replayable game, and should get return traffic a bit better than Broforce did.
It's such an outlier in terms of hype and expectations. The fame of Braid is about as good as it gets in the indie scene, and besides that, indie first person puzzlers have a growing audience (and the Witness by all accounts seems to be a good one).
It's a thumbsuck, but on PC I'd guess 270,000 in the first two months. I'd expect it to reach 600,000 in a year's time. I expect Jonathan Blow to be slow on the discount uptake, otherwise I'd expect more sales within a year.
I'm going for combined platform in this, reasoning:
Sony has put a lot of marketing into the release of the game, even if it has more success on PC they created a sense of mindshare and wonder about the game with their marketing.
Braid is an institution in the videogame space, perhaps less so now that "the indie bubble has burst" but I think that "The Creator of Braid" will add wind to the sails
The detractor is that the game is retailing for $40. So I'm not really sure what will do to the fiscal potential of the game.
Given the above I think it will sell 350k copies in the first 3 months, and break 800k in a year given that we won't see more than a 25% discount on the game in that time.
Because of the price tag I opted to base my notion more around a $ return (1mil in 3 months) and made as close a pretty looking number as I could.
In all honesty I'm now 100% certain about the year estimate, and far less analysis went into that supposition than the 3 months.
I feel that my estimate is conservative while not being pessimistic
On the Witness:
Totally awesome that it's being priced at $ 40. I suspect that the Witness' sales will largely be driven by the massive media coverage around launch and people's memory of Braid. I think this will result in slightly higher sales than if it had been launched by someone that isn't Jon Blow and the Steam reviews might also suffer a little (cause more people buy it instead of "just" those that would love it).
So my estimate on Steam sales is: 250 000 in 3 months and a 95% rating (I think the rating would be higher if not for the hype). I looked at Talos Principle for comparison.
It would be awesome if this was some sort of app that asked you for an estimate at a relevant time and then reported back on how accurate you are after a duration. Doesn't sound like it would be too difficult to make... hm.
Personally I think the game is going to do well. The game has been seeing a ton of exposure and Firaxis's willingness to send out review copies a month before launch make me think they are quite confident in their execution for the sequel.
I'm going to estimate 750,000 in the first 3 months. I think it might do more but I'm worried that's my own overeagerness speaking. I'm basing that number on the fact that XCOM: Enemy Unknown had more than 200,000 players in the last two weeks, as well as having more than 2,5 million lifetime players.
This Friday 11pm it so happens he will be looking at Cadence. ;)
http://www.twitch.tv/ryan_clark/profile
I know Super Hot is pretty rad, and pretty novel, having played the jam version. And Super Hot had that successful Kickstarter, and they've been working on it for a while. So I expect it to do very well in reviews as well as word of mouth as well as pretty well in Let's Plays (for the launch period, I don't think it's replayable).
I think the game might get quite a few core gamers playing it. And it's going to get a big proportion of the indie gamers who also like FPS's. I don't expect there's going to be a bigger indie FPS this year. I think that there is a fairly big overlap of indie gamers who play FPS, but it's not as big as the overlap of indie gamers and First Person Puzzlers, and although Super Hot is as much puzzler as shooter, I think it does appear to require previous FPS skills to really enjoy.
So I'm going to predict 240,000 in the first two months (about 3 times that of Broforce). And 600,000 in the first year. This would mean it is an upper tier indie success. Not very indie games by small teams do better than that. If the play time is low, and if there isn't much to it besides a series of puzzles, then I might be overestimating.
Because if it's the latter, then the 2 months have already passed, and there's already data on that. SteamSpy seems to show that as of 5th February, 2 months after launch, there were somewhere around 53,700 more owners of the game than before.
While I'm there, I can't view that far back to see the 2 months since launch exactly, Galak-Z but it seems like it sold around 9,000 copies on Steam 2 months after launch, and 14,700 copies up til today.
I was optimistic, at 80,000 sales :(
We were all very optimistic about Galak-Z. I was the closest with 25K sales (nearly 3 time what it made). But in our defense it looks like Galak-Z was much less loved on Steam than it was on PS$, it's sitting at a 82% positive review rating, which nearly puts it in the bottom half of all games on Steam by review score.
Galak-Z has pretty great production values for an Indie Game, but Steam users don't think it has particularly great value in terms of an experience (which perhaps shows you can't production value your game into being successful, past a point anyway). Reviewers overall gave it 82% on Metacritic, which quite likely shows that critics didn't bother finishing the game, and gave the game the benefit of the doubt based on the history of the developer and the production values.
Galak-Z probably didn't do well in Let's Plays, given the sales. Which I guess means that sort of being a Rogue-like isn't enough to interest Youtubers (there's likely plenty of Rogue-likes to play these days).
It's a game that IMO lacks a strong hook/doesn't have the strongest theming, but the execution and polish is at a high level and it has a good possibility of being picked up by streamers and long form youtubers. It'll also be called a nuclear throne clone for better or worse. Still, lack of a strong hook, and the roguelite genre might be past its prime, so I'm going to guess 30k in 3 months, but I won't be too surprised if it greatly surpasses my expectation.
I don't like the theme of Enter The Gungeon at all, not when compared to Binding Of Isaac or Nuclear Throne. I think it looks cute, but not particularly charming. I'm not sure whether this will affect it's adoption with the Youtube audience, being a new rogue-like, and a fairly polished one, and published by Devolver. So I expect people to adopt it quite quickly, and after a few plays be hooked, or put it down. My instinct is that the theme makes it a difficult sell despite the apparent variety of the gameplay, but I think that young players may like the theming more than me (and that my taste is causing me to under value it). I'm going to guess 32,000 in the first two months (a bit less than Nuclear Throne), and 130,000 in a year.
[Edit: After seeing the launch trailer I think I may have underestimated]
http://www.gamasutra.com/blogs/LarsDoucet/20170509/297708/What_I_learned_playing_quotSteamProphetquot.php
There's a few things he's doing better though, he's getting a list of upcoming games from Steam (rather than cherry picking games he's already heard of) and they're ranking the games (which I think might be simpler than trying to predict sales for each). Also they're doing a calculation of earnings rather than sales (which makes the performance of $3 games and $30 games comparable).
Anyone keen to do this? I'll probably be doing this internally anyway, as I think it's a crucial exercise if you're going to be making huge time commitments based on predictions about your own games' sales.