The Valve and Steam Platform Discussion Thread

[KVVRR]

lucky thing that they still put $40 games to $5 for whatever reason here

 

[Jenkem]

this is old as fuck, this happened about a year ago. no idea why they are just talking about it.

 

[Jenkem]

Danikas explain yourself faggot

 

[Duraframe300]

Question: How accurate is Steamspy really? Especially their *owners* number make no sense in some cases. I recently noticed it again with Grounded which has between 50'000 and 100'000 owners while Destroy All Humans is between 100'000 and 200'000.

This makes no sense as they both came out the same day and Grounded has been above DAH since release.
Even now Grounded is at 3rd and Destroy All Humans barely makes the Top 20.

I've seen this with other games as well. How do they estimate those numbers?

 

[Infinitron]

Steamspy became a guesstimate over two years ago, after Valve shut down the method it was using to get accurate numbers.

 

[Baron Dupek]

https://www.kotaku.com.au/2020/08/s...y-removed-some-games-purchases-from-accounts/

Last night a temporary glitch hit some Steam users in the worst way possible, removing games and additional purchases like DLC from some player’s libraries.

It affected all kinds of games, from Red Dead Redemption 2 to Fallout 4, with the only common trait seemingly being that they were all titles that were on sale for one reason for another.

Users would try to find games in their library and…they were gone. And if they tried visiting that game’s store page to launch from there, they’d be prompted to buy the game they’d already bought.

The good news is that the issue was resolved in a few hours, and all that was required was a Steam reboot:
The bad news is that, holy shit, it is very easy to lose access to things that have cost you hundreds if not thousands of dollars when you’re not holding the actual game in your hand.

 

[BlackAdderBG]

"...hit some Steam users in the worst way possible...It affected all kinds of games, ...Fallout 4"

 

[PulsatingBrain]

Is there a way to see if a game or DLC has been on sale before?

I'm playing American Truck Sim and I'm wondering if waiting around for a sale on the DLCs is realistic, or if I should just get what I want now

 

[Chimera]

PulsatingBrain

https://isthereanydeal.com/game/americantrucksimulator/info/

Basically the go-to site for pricing where digital games are concerned. Mind, they only collate data from direct-publishing and approved vendors, never key-resellers (gray market or otherwise).

 

[Alienman]

There is also this site:

https://gg.deals/game/american-truck-simulator/

It also lists key shops but with a disclaimer.

 

[DalekFlay]

The bad news is that, holy shit, it is very easy to lose access to things that have cost you hundreds if not thousands of dollars when you’re not holding the actual game in your hand.

I love how everyone once in a while someone loses some movies or games and the media suddenly realizes the downsides of DRM for a split second, then goes back to not giving a shit the next day.

 

[Clockwork Knight]

reports of entitlement issues for Steam users

Entitled gamers at it again!

 

[Infinitron]

A new heuristic for estimating game sales based on number of reviews: https://gamediscoverability.substack.com/p/how-that-game-sold-on-steam-using

How that game sold on Steam, using the 'NB number'.
Graphs are included, please keep up at home.

[Hi, I’m ‘how people find your game’ expert Simon Carless, and you’re reading the Game Discoverability Now! newsletter, a regular look at how people discover and buy video games in the 2020s.]

So, here we go - it’s the results of this newsletter’s gigantic dev survey on how many sales per review Steam games have. And before I present the results, just wanted to make sure we were on the same page about why it’s important, and what the numbers mean.

Why do we care about sales/reviews ratios?
It goes like this: we would love to know roughly how many copies of games are sold on Steam, for comparison and planning reasons when we make games. That data is not public. But the number of player reviews is public. So historically, there’s been the Boxleiter number, most recently updated by Jake Birkett in 2018.

So what I’m going to call these new results is ‘the NB number’, or New Boxleiter number (hi Mike Boxleiter, whom this is named after, btw!) And it turns out these numbers - we now have data per year - have indeed been changing quite a bit.

As I noted in my original call for data, Steam introduced a ‘would you like to review this?’ button (shown below) at the end of October 2019. This turns up sometimes when you’ve played a game for more than 2-3 hours and haven’t reviewed it yet:


This has led to lots more reviews of older games. And has definitely changed things for games launching after this UI tweak. So let’s get to it.

Introducing.. the actual ‘NB number’ dataset!
Starting out, here’s a sweep of all the data I got. Note that I did remove a handful (10-ish) of outlying data points - mainly from F2P games that had very high review/sales ratios because they were free to download, and a couple of incomplete submissions.


So over the 237 (!) non-outlying Steam games that submitted data to us, it’s an ‘NB number’ average of 63 sales per review, and a median of 58 sales per review. Thus, if you’re looking at a Steam game of indeterminate age, that’s the kind of lens you can put on it.

The middle 80% of the sample is roughly between 25 and 100 sales per review, so that’s a good range to consider, if you’re not looking at year of launch so closely.

(BTW, if we go back to Jake Birkett’s survey data from 2018, the numbers were quite a bit higher - 82 sales per review as an average, and a median of 77 sales per review. So it’s definitely been coming down, as more people are encouraged to review games by Steam - and even before then.)

Spoiler

Regarding year of launch, we can see that there’s a pretty large difference in average/median sales per review, depending on when a game launched:


So if you launched your game before 2017, you still have an ‘NB number’ average of 81 sales per review and a median of 74 sales per review. But if you launched your game in 2019, when that significant change occurred, you have an ‘NB number’ average of 57 sales per review and a median of 51 sales per review.

And most pertinently, if you launched your game in 2020, you have an ‘NB number’ average of 41 sales per review and a median of 38 sales per review. And it seems like the current non-outlier range is between 20 and 60 sales per review - for a game launching from scratch this year.

Since many of us care about the present day, here’s the data set of ‘sales per review’ purely for games launching in 2020:


In many ways, don’t forget, more reviews is good for games, since a prospective player can see more feedback. BTW, the average Steam review score of those responding was 84% positive, and the median was 89% positive.

There was no major review positivity difference I could see after the ‘review prompt’ was added. I may put out some bonus graphs around review score and copies sold in subsequent newsletters, since there were a couple of interesting trends to highlight.

Before and after review prompts…

Just to reinforce why looking at review/sales ratios per year is important, Erik Rydeman from Clone Drone in the Danger Zone dev Doborog pinged me as I was prepping this newsletter, and said:

“I think you're absolutely right to take the changes with Steam encouraging reviews into account. Our lifetime sale/review ratio is 30, but looking at the range from only this year it's 14. Looking at 2017-2019 it's 67. Spot the point where Steam changed their UI!”


And wow, he’s right! I checked some other games that I have access to sales data on, and I think Clone Drone is a bit of an outlier.

It’s great, has repeat players, and thus had a lot of folks who had never got round to reviewing it, but were still playing. Nonetheless, Erik’s game is ‘catching up’ on reviews from people who bought a long time ago, thus the super low NB number in 2020 alone.

But the effect can be there, if less pronounced, with most Steam games that had been live for a while by October 2019. (Some exceptions exist - if your game is ‘one and done’ and older purchasers played through it once, even a high selling title won’t have got extra reviews from the reminders. New players will still get prompted, of course.)

Conclusion: do your math(s) right!
Anyhow, that’s your data! Hope you find it helpful. Before we conclude, just a reminder, in the classic ‘jumping to conclusions about if devs are millionaires or not’ fashion.

If you think a game has sold 20,000 copies using the NB number, and you want to work out how much money that title made, please don’t multiply that by its regular Steam USD store price and think that’s the money received in the dev’s pocket.

You’ve got to factor in average ‘regular’ price on Steam, how much and how often the game was on sale, and then after the money comes in, Steam platform cut, refunds, and VAT before you even get close to working out the real number. (Realistically, that’s 30%-50% of the ‘optimistic’ gross number, and potentially way less, if discounting is aggressive.)

Video games, eh? But we still all like playing them.

Take care until next time,
Simon.

 

[Country_Gravy]

I appreciate the posts of Infinitron in this thread.

I quite literally get all of my news about the Steam platform from here.

 

[LESS T_T]

 

[cosmicray]

Valve continuing to dump money into Linux gaming: https://www.gamingonlinux.com/2020/...linux-graphics-drivers-starting-with-amd-radv

I wouldn't call it "dumping", it's just one developer. I assume they are "losing" money in doing this, at least for now, but it's a long-term game. And they aren't dumping unreasonable amounts of money into it. Plus they might profit from people on Linux gaming/buying more. I'm certainly one of those.

 

[Horvatii]

Conclusion: do your math(s) right!
Anyhow, that’s your data! Hope you find it helpful.

Well, acshually there is no conclusion and the data is not helpful... cause there is absolutely no use in the avr/median nr of sales per review to anyone but someone betting on that specific number... They have almost a unifrom distribution and give no variance/standard deviation to it, while it would be gigantic. So what is anyone supposed to do with the avr/median? You cant even properly guess the nr of sales of it... you could be far left, you could be far right - and have no way to correct for it.

For a game dev there is no use since you cant really release a lot of games and hope for an average...
To an investor its meaningless.
Revios/sale are going down Steam-wide year on year, thats the only meaningful result, but what to do with it... what does going down mean, what does Steam-wide mean... and on and on

Like, its not in any way normal (black), so you have zero confidence in a guess.
Its not even uniform, so you have decent guess left or right.
Its actually non linear too, so you have ZERO parameters you could potentially use to correct for. And when you guess, you could be further off, than expected cause its not uniform.
Avr/median is a trash number here, that is actively misleading if used on individual titles.

 

[Horvatii]

Like, someone reveal an idea on how to use this "NR number"?
Stuff like 'sales are between 20k and 80k' doesnt count.

 

[mk0]

You don't, this data is only relevant in the context of comparing it to the data from the previous years. What it's telling you is that the supply for games outstrips demand on Steam and that this gap has only kept increasing.

 

[OSK]

Valve continuing to dump money into Linux gaming: https://www.gamingonlinux.com/2020/...linux-graphics-drivers-starting-with-amd-radv

I wouldn't call it "dumping", it's just one developer. I assume they are "losing" money in doing this, at least for now, but it's a long-term game. And they aren't dumping unreasonable amounts of money into it. Plus they might profit from people on Linux gaming/buying more. I'm certainly one of those.

It's one more developer. In total, they've put quite a bit of money into Linux gaming. From the article:

Considering all the resources Valve are putting into Linux gaming across a number of developers to work on the actual graphics drivers, the ACO shader compiler, the Steam client on Linux, the Linux Steam Runtime container system, working with CodeWeavers on the Proton compatibility layer for Steam Play and more they must be pretty confident in their plans for Linux gaming as a whole. No matter what, everyone on Linux ends up benefiting from all their work since it's largely open source.

Their entire investment might only amount to several million dollars, chump change to Valve, but that's a lot for open source projects. Even more so for contributing specifically to Linux gaming. Epic donated $25k to Lutris and that made news.

 

[Metro]

I appreciate the posts of Infinitron in this thread.

I quite literally get all of my news about the Steam platform from here.

Only Steam news that would interest me is Valve actually making a new Half-Life (non-VR gimmicky bullshit) or L4D.

 

[Immortal]

L4D.

But why.. it's the same recycled shit..
You would really be hyped for next-gen Zombie graphics and a new cast of random fucks?

 

[Horvatii]

What it's telling you is that the supply for games outstrips demand on Steam and that this gap has only kept increasing.

Does it though?

 

[BlackAdderBG]

What it's telling you is that the supply for games outstrips demand on Steam and that this gap has only kept increasing.

Does it though?

The supply of shit is thru the roof, no wonder the demand is lacking.

 

[Silly Germans]

I really dislike it when people claim that Valve saved PC gaming. Its like saying that Amazon saved reading. I couldn't care less if PC gaming is a profitable business, which is the only thing that Valve "saved". I found pc gaming much more enjoyable when there was a handful of releases a year made by enthusiasts.