HERE’S A LONG ARTICLE EXPLAINING WHY YOU MAYBE SHOULDN’T USE KEY DISTRIBUTION SERVICES
One indie game developer's experience with key distribution services outlined
[N.B. This was originally posted back in August 5th, 2022, as a Google Drive document. I have reposted it here with some light edits to be more permanently available.]
INTRODUCTION
For the past three months we’ve been using a product key distribution service that, surprising no one, ended up being an informative waste of money. This is a fair enough outcome considering we had no expectations or explicit goals. Mainly, we wanted to get more eyeballs on our game as the various summer sales of 2022 started to kick in, and we had little to no experience interacting with “new” streamers or content creators we weren’t already familiar with – so that’s the justification for doing any of this. Whether any of it moved the needle, we don’t know and we won’t be repeating this.
Importantly, in the author’s opinion it is not key distribution services that are ultimately at fault – the actual problems are baked into the platforms of YouTube and Twitch that they rely on – so we’re entirely reticent to either directly name the service or any of the people involved. Twitch and YouTube are the ones that allow such accounts to persist that in turn have created these cottage industries of discoverability, and incentivize subsequent behaviors from both content creators and obvious scammers. Unless the platform holders take steps to address this sort of abuse of their systems, this problem won’t go away. While we were aware of bogus or outright fraudulent accounts on YouTube and Twitch, we weren’t quite prepared for the scale of how widespread it was. For perspective, we’re a small indie studio with just one game out and hardly anywhere near “big”, but even then the amount of obviously dodgy accounts to wade through far exceeded our expectations.
All that said, there are some things to this experience that were useful (hence the “informative” part), which we’ll elaborate on, but there are much better ways to spend one’s money – in our case it probably should have been on ads. Lesson learned.
BRIEF EXPLANATION
So what is this key distribution service? It is intended as a go-between for publishers/developers and content creators to distribute product activation keys. Simply put, either you as the developer or publisher turn up with keys for your game, and the service helps you distribute those keys to content creators that have signed up to said service. While the service claims to screen content creators, after minor investigation it is extremely easy to abuse access to the service through things like paid-for CCVs and views.
In terms of money spent on our end, it was under 600 USD for about three months, which allowed us to set up campaigns in the service’s backend, and usage of what they call their “insights” tools, as well as support via chat when needed. As a first-time user they also offered to feature us for two separate weeks on their campaigns “featured” section, which under their usual offer is a type of paid placement to highlight campaigns to content creators on the platform.
Campaigns are how you distribute keys. You set up a start and end date, you assign a number of keys and set the conditions for taking part in the campaign. Content creators that meet whatever requirements you set get to request a key. Key distribution can be automatic approval or manual – we went with the latter as we didn’t want several dozen keys to evaporate overnight.
The insights tools are arguably the most useful thing about the service but infuriatingly have little to no documentation associated with them. These tools are things that access the Twitch and YouTube APIs to retrieve account information available in said APIs. So, such tools allow you to do things like easily retrieve a list of content creators, sort them by a particular category, language spoken, and other typical metrics like follower/view count. For example, if you’re wanting to look for people who have a history of playing games like the one you’re making, you can put a bunch of search operators in to hone in on those particular creators. There’s also a listening tool, which will search things like YouTube results and gaming websites for mentions of whatever you’re looking for. Useful to have all in one place, but other free sites like sullygnome and tools like Google Alerts already exist.
Support via chat was also fine – in fact it was only through them that we were able to make some use out of the insights tools. For instance, considering our game is localized into 11 different languages other than English, we could use these tools to quickly find a list of content creators that have played games similar to ours but were not English speakers and invite them to the campaign, even if they weren’t on the service itself. It is highly doubtful any of these emails were successful and we didn’t receive any explicit notification of them taking up our offer. Our own prior experience interacting with other content creators is precisely that: they vastly prefer to speak to or receive an email from an actual person than have some generated message land in their inbox inviting them to a faceless promotion for a service they’ve probably seen several different forms of already.
CAMPAIGN DESCRIPTION
Here's what our campaign consisted of:
Start date: May 10th; end date: August 18th
100 Steam keys for deluxe edition of game
Requirements of either 5 avg CCV (Twitch) and/or 75 avg views per video (YouTube)
Manual approval of key distribution
Campaign “featured”: May 23rd to May 29th & June 20th to 26th
Other optional things provided: various PNGs of game capsule art and logo, UTM link for Steam page, media kit containing b-roll and other associated game assets
Other factors to consider were two summer sales (GOG in late May and Steam in June) that fell approximately on the dates our campaign would be featured in the service. Any impact this service may have had on either sale is difficult to determine. As one would expect from looking at a sales graph for this time period, spikes would occur in time with the sales. Though before we get ahead of ourselves, we should look at what this campaign was directly generating: content.
RESULTS OF A KIND
So far out of the 100 keys in the campaign:
74 people have claimed a key after being manually approved (once approved, the content creator can then claim their key).
Of those 74, 38 have submitted content – 11 YouTube videos and 36 Twitch streams – while the remaining 36 content creators have not submitted any content tagged as our game.
At least 67 accounts that applied for a key have been manually rejected.
We’ll elaborate on the rejected accounts and how/why accounts were approved later but that number was far higher than expected (it’s also slightly higher than 67, because after reporting some obviously fraudulent accounts to the service, the service removed them entirely, skewing the number in the “rejected” pile). Again, we are not a huge game, it’s a single player title, we’re in the top ~8000 sellers on Steam after six years – so we consider ourselves somewhat niche. Even sitting in that niche, just over half of the requests were deemed “legit”, and just over half of those submitted content after receiving a game key. Or, numerically, 38 out of 141 apparent content creators, or ~27% produced something (the quality of which is debatable). The batting average here is not great but as said in the introduction: don’t do any of this.
Speaking to the point of “producing something” – the implicit trade off of content creators continuing to get free access to this key distribution service is that they at minimum make some sort of content tagged as our game (or any other game they might claim a key for) or risk being booted from the platform. Our goal was at best eyeballs and word of mouth to creators from other countries who otherwise might not have heard of us – if we really wanted to get a slew of videos made there are other much more direct methods to do so.
Anecdotally speaking, because we don’t have a graph of this, as expected during the two weeks of being featured, requests for keys spiked.
A note for the next part: there was some amount of overlap because content creators will sometimes cross-post Twitch VODs to their respective YouTube channels rather than archive them on Twitch. In addition, at least one streamer streamed on two separate days. In other words, while there were 38 content creators that submitted content, in total about 47 VODs and YouTube video uploads were produced. Also, we are focusing on airtime/duration as the most significant metric rather than CCVs/recurrent viewers or the creator’s follower count considering the overwhelming majority of videos were effectively “one-shots”.
In terms of the 36 Twitch streams produced, 13 of them streamed the game for an hour or more. Fourteen streamers played the game between an hour and “0.2 hours”. Due to how Twitch’s API apparently works, it does not register a stream’s length shorter than “0.2 hours” – or in other words 12 minutes or less. There were nine such sub-12-minute streams, where the worst instance ran for approximately 35 seconds and the content creator did not even display footage of the game – just long enough for it to apparently be registered in Twitch’s backend and thus also registered as “submitted content” with the service . At the other end of the scale, one streamer with only a handful of viewers streamed the game twice, first for almost an hour as a test to check audio and video, then played another three hours live. Probably the “best” performance was a content creator with 44 peak viewers and 37 average viewers who streamed the game for two and a half hours. In aggregate of time of writing (August 2nd, 2022) the service produced the following graphs for Twitch streams.
There were 11 YouTube videos produced. The longest had a duration of 59 minutes and 35 seconds, the shortest was 7 minutes and 3 seconds. Technically there was a longer video of just over two hours, but the majority of that video was of an entirely different game with the remaining ~20 minutes devoted to our game. YouTube’s API apparently doesn’t operate in the same way as Twitch’s “chapters” do to properly demarcate the amount of time for footage of a certain thing. So, for example one could upload a video to YouTube that’s 7 hours and 55 minutes of Minecraft footage, put five minutes of your game in, and in the eyes of the key distribution service you would think that video was an 8-hour upload of your game. Somewhat shockingly, one account, upon receiving approval for a key, rapidly turned around an 8-minute “review” video the same day of which the first 90 seconds was an ad for some sort of monetized mobile app. In aggregate, the service produced the following graphs for the YouTube videos “submitted”.
Even with so few videos, just over a thousand total views is not even a drop in the YouTube ocean.
DID THIS MOVE THE NEEDLE AT ALL?
The short answer is we don’t know – and if it did, we didn’t notice. Remember that there was a UTM-link for our Steam game? This is how well that performed.
Irrespective of thousands of minutes watched, the best it did was 55 visits, 12 of which were tracked, that resulted in zilch. It is entirely possible viewers may have gone on to wishlist or even purchase the game separately and not bothered clicking the provided UTM-link, but there is no way of proving that in this situation.
The longer answer is, judging by Twitch’s own analytics it provides to developers (which seem to be unreliable if you don’t reach a minimum threshold of hours watched), there wasn’t any noticeable change in the usual amount of streaming. Hours watched on a weekly basis can be anywhere between single digits to around 100 hours depending on whether we have a weekly deal occurring. Meanwhile, to see if more people were actually playing, we’d look at SteamDB. It will typically show us around 20-40 concurrent players over a week. For the weeks when the key distribution service would feature us, we didn’t see much of a deviation from the usual norm of the past couple of years. In fact, when there was an occasional spike in hours watched or concurrent players, it seemed to be coming from other streamers playing the game who had absolutely nothing to do with the key distribution service and ended up drawing way more eyeballs alone than the content creators from the service.
SHUT UP AND SHOW ME SALES DATA
On GOG for the entire period of using the service as of time of writing (May 10th to August 2nd, 2022) the units shifted looked like this. There was also a summer sale at the end of May.
Given that we were only providing Steam keys and have no form of tracking click-throughs from the service to GOG, it is very unlikely it moved the needle on GOG. Also, given the typical increase in traffic and units sold caused by sales, any benefit brought about by our use of the key distribution service is not possible to discern. In fact, looking at one of the periods where our campaign was featured in the key distribution service that didn’t overlap with a sale period is unremarkable.
Over on Steam the sales graph for the same period spikes twice:
The first was for a weekly sale, the second larger spike is for the two-week summer sale. Once again, the first featuring period – which comes right after a weekly sale – is unremarkable and didn’t deviate from what we usually see, while the second featuring period overlapped with the 2022 summer sale.
Given that the UTM tag turned up neither wishlists, purchases or activations – the conclusion to be made here is that this key distribution service did not positively impact sales data.
OKAY SO WHAT DID YOU LEARN?
As stated in the introduction, these problems are baked into both Twitch and YouTube and until either platform changes how it operates, short of manually investigating the reputability of each account seeking a key (which is what we did) you are exposing yourself to very large numbers of bogus content creators and streamers.
Interestingly, a lot of common features started to become noticeable about the obviously bogus accounts – here are some things to watch out for specifically on YouTube:
Avatars that look like they were made for cheap fictional baseball teams
Channels that are “dead” or might have had one video that did very well years ago but do not appear to have uploaded much since – such accounts are seemingly abusing their “record” on YouTube to get easy access to key distribution services.
YouTube channels created on or around August 19th 2015 – YouTube/Google apparently made some change to how it handles accounts on this date and it’s apparent this was abused to generate multiple channels all pulling the same scam
No business email in the About section
Channels that mysteriously have the same count for both subscribers and views on a handful of videos published barely a week ago – these are obviously bought subs & views
Videos that are just 10-20 minutes of gameplay footage and do not feature any sort of overlay or person on screen
While on Twitch, watch out for the following:
Avatars that look like stolen EverQuest/World of Warcraft fan art
Follower count means nothing, but the typical scammer number seems to be anywhere between 2k-12k followers
No VODs published or recent stream visible – Twitch by default doesn’t publish VODs and if they are they expire after 14-30 days but can be “highlighted” to stay up forever
If a recent stream is visible, but no streamer/overlay is visible or talking over the gameplay, it’s probably a phony account
Activity of one stream a month or less and/or nothing visible in the Schedule tab
Accounts that apparently play a one-off 12-hour marathon of several different games across multiple genres and then nothing else ever again before or after – some of the games listed below are unfortunately associated with such Twitch accounts (possibly due to being part of free giveaways or bundles in the past)
Any account in its About section where it’s shilling energy supplement crap that looks like this
Not every account that was rejected was bogus. Other accounts that seemed to be legit might have been rejected because, based on their activity, were perceived as being too far from our intended audience – but that is unlikely to improve your chances. As pointed out earlier, some of the accounts approved would either not stream at all, or if they did, the content produced too often felt “forced”.
On the upside, when some approved accounts claimed keys, they often advertised this on Twitter in advance that they would be streaming. This allowed us as developers to occasionally drop in on such streamers and join their chat and generally have a merry time. It’s no surprise that content creators tend to be enthused when people turn up to watch them play, more so if it’s with the people involved in making the game they’re playing. In addition, a few of the content creators seemed to make some use of the PNGs provided in the media kit for thumbnails.
SO WHAT’S THE CONCLUSION?
All of the above can be done organically with existing tools and other free services, but as a developer it will still require your time, energy and money to do so. In our opinion key distribution services like the one we used do little to facilitate this process.
The chief reason why is when it comes to content creators and streamers – as cliched as it sounds – above everything else they value their authenticity (and so does their audience). Even when they use such services, they are still careful to make the fact known that the product key was not purchased and as such would label the stream or video with some equivalent of hashtag #ad. With this sort of labelling, the type of content produced is already going to be skewed in a particular way and no amount of smoke and mirrors will mask this, because an audience will see through it.
Content creators are people and typically wish to be treated as such. Key distribution services like the one we used do not facilitate any form of direct communication with content creators on their platform – potentially for good reason – but all this does is add more hurdles to the long game of the “influencer relations” process. Building a rapport with content creators, streamers, youtubers or anyone who posts any form of content online takes time and trust – it is not an instant process and should not be treated as such.
TL;DR – if you’re going to spend money on promoting a game, you might want to spend it on social media ads instead of key distribution services – particularly if you aren’t interested in spending time.