Head of User Acquisition Playkey Ruslan Sharapov has published an extensive story about the Playkey company Twitch channel. We would like to share its story, highlights and results. Enjoy the translation published in our blog.
A systematic approach to user acquisition is all about working with hypotheses. Sometimes, the conventional ways of getting traffic (FB, Google ads, etc) just are not doing enough.
Just a few days ago we at Playkey summarised the results of one of the longest-running hypotheses. It took me 26 months, that is just a little bit over 2 years.
The Hypothesis: we could make a Twitch channel and acquire users at a cost lower or comparable to other sources. The big idea was to grow the channel big enough, so we would not have to spend money buying promotions from other streamers and would just generate users from the Twitch channel organically.
In case you prefer numbers, this is what it’d look like:
- CPU =< $0,33
- CR >= 2,5%
Spoiler: one of these numbers ended up being 4 times better, the other one was 14 times below our expectations.
It was clear from the start that this was a long game to play. It normally takes a year and a half of regular streaming for a content creator to grow a sizable audience. Yet, I had a plan. Why not use Playkey resources, with tens of thousands of paying users, over a hundred thousands of social media followers and an enormous email listing? It seemed like you could grow quickly. Just use banners on the main page, browser and client-app push notifications and lots of emails. good game, well played?
Spoiler: you wish.
Gotta keep it SFW!
I also had a goal of not making it a Just Chatting stream. You know, the type of content when you get a pretty looking female just talking to subscribers. Sometimes they are not even playing the games, just chatting, occasionally leaning towards a camera or picking up something from the floor. This would definitely give the channel a good number of subs, yet it is very unlikely that the horny teens would even click on anything, let alone pay for a cloud gaming service. As a matter of fact, Twitch has been trying to fight the booby streamers for a while now. Though it seems like they are not doing particularly great. Even when they are trying to enforce the guidelines, the new booby streamers appear literally the next day. Just Chatting remains the most viewed category of content on the platform.
You have to understand that Twitch is generally unwelcoming for new streamers. There are literally no ways of boosting for beginners. So, unless you are somebody in the world of cybersport or a female willing to flash some underwear, you are gonna have a bad time. Be prepared to stream for 1–5 people for at least the first six months or save up a considerable promotion fund. There are practically no chances of making it in the Twitch game without spending a whole bunch on ads, no matter how original your content is or who hosts your channel.
Twitch will demotivate you in every possible way: the main page will have verified partners all over the place, in categories you’ll be impossible to find even if you know where to look for. They’ll easily slap a ban on you for a week or so for the tiniest of screw-ups, no matter if it’s a report from a salty kid or a copyrighted ringtone accidentally played on your phone.
The channel was run by me (production, traffic-management, sometimes directing and writing), a charming young guy who’s extremely good at shooters, a cute-looking cosplayer girl and a wise panda (literally an employee wearing a panda costume and streaming at his spare time).
How did we find the streamers? There was no job posting or anything, I just asked around from friends and colleagues and after checking out the streamers settled on the two guys. A combination of some experience in streaming games and rather small viewer numbers (~30 people) made them the perfect match. The initial offer was $266,8 per month with 70% of all donations going to the streamers. Naturally, the payment would grow along with the channel and its KPI.
I am deliberately not giving any opinion on the quality of the content. That is a subjective thing and there is no way of telling what streamer is gonna blow up or who is stuck streaming for 50 people for years. I did try to figure out the secret sauce, I parched some info from Twitch’s API, talked to other streamers and was watching social blade. That did not get me any closer to figuring out the key to success on Twitch.
I should probably still mention that we did not just stream, we tried AMA-sessions with the devs, educational streams, interactive stuff like something you’d hear on the radio and even news/reviews for anything relevant to the cinema/movies/comics and even talk shows featuring several streamers.
How did we promote the channel?
- We streamed without branding. For the first two months, Yuna was pretty much the face of the channel. Twitch viewers are not too happy about corporate channels, so the test streams and initial channel growth was done with no branding.
- Channel Points system. Set up a chatbot and launched a stream store. Viewers would get points for just watching the stream or chatting, the points could be exchanged for all sorts of goodies including free gaming time on Playkey.
- Bombarding Playkey audience. Banners on the main page, regular mentions on the social media channels, email, embed-code with the window directly on our website, push notifications in the app and browsers. We did get a few thousand viewers, though they did burn out in just a couple of months. We tried reactivating them with different marketing materials and even offers, but it didn’t really work.
- Hosting the Twitch stream to our VK group (over 60k members) and YouTube channel (almost 9k subscribers) with chat being joined from all three services. No luck, while YouTube did show at least some activity, VK subscribers did not respond at all. We stopped hosting streams after 3 weeks.
- We made a landing page with a bio for each streamer, streaming schedule and the goodies people could get for watching the streams. Brought traffic from various sources to the page. Can’t really measure the practical result, but I get a sense it wasn’t much.
- Google Ads. YouTube with targeting on specific channels and Google Display Network. Ended up being too expensive.
- FB Ads. Could not get a decent CTR, had no resources for a proper campaign with DCO.
- Tried VK Ads. This is a whole story of its own. We launched the first ads and everything was going pretty well, CPM and CPC were good until VK mods shut it all down a couple of days later. Tried relaunching the campaign. This time couldn’t even go past pre-moderation. I contacted the support team, and it turned out that Twitch is a direct VK competitor, so any ads linking to that domain would be blocked. They would make an exception for us, but that would require a $26 680 per month ad budget. Naturally, I had to politely decline that offer.
- We tried buying viewers from Reyden. It’s a specialized ad network for streamers. You give them a link to your channel and as soon as you start streaming it starts embedding the live video to other websites. So, people would see the stream as an ad on torrent trackers, video streaming sites and gaming media resources. They provide detailed analytics, so the results are transparent and fair. Looks good on paper, way better than just straight up botting, right? There is a catch, though. Twitch does not count embedded views while considering a partner program application. Viewers on third-party sites are not always aware that this is a stream and not a pre-recorded video and they just want the ad to be over asap, so they can go back to content. This tool works, but Reyden alone was not doing enough for the channel.
- We tried buying ads from other Twitch streamers via StreamPub. That’s a specialized agency and a service that helps to buy traffic from streamers with the emphasis on motivational promos (giveaways, prize draws, etc). Didn’t really give any meaningful results to Playkey or the channel. Perhaps, we picked the wrong streamers, or the content format wasn’t right. Either way, we moved on and did not investigate deeper. The service was relatively expensive and required too much fine-tuning.
- Bought ads directly from smaller YouTubers (up to 30–40k subs) with high ERV and ViewsRate. That actually works but takes time to pick the right streamers and negotiate. Smaller bloggers are not yet experienced enough when it comes to dealing with advertisers, they forget to check emails or need a lot of legal paperwork. Some are just straight-up unreliable and can disappear for a week or two. Agencies are not willing to deal with them, the size and commissions they’d make are not worth it.
- Collabs with larger streamers. It seemed to give some boost, but the results were hard to measure. It would require regular time investments and patience to talk to streamers, find new ones and sometimes even take rejections or deal with being ignored altogether.
- Created a Coub channel and uploaded short coubs with fun moments from streams. Most of the videos would get 1–2k views, though there were a few that had 60k+. Thanks to that the content also appeared on YouTube and Instagram and got a few hundred thousand extra views there. It’s hard to estimate the impact, but I would say it wasn’t significant.
- We occasionally uploaded stream highlights to YouTube. The best we could get was a few thousand views, none of the highlights went viral.
- Host fan tournaments with prizes. A Playkey Devs vs Users cup was quite successful. Works pretty well as a way of increasing the loyalty of the existing users. Though it does not bring new users.
- Live streams from festivals. New experience and a great way of interaction with subscribers. Brings high quality content, yet is not particularly effective for bringing new users.
What about the numbers?
- 1 556 387 views of video content on Twitch
- 5 055 hours of streaming (over 6 hours of streaming per day, no days off)
- 26.9M minutes of watch time
- 947 757 unique viewers (including pre-rolls from Reyden)
- 34 069 followers (including 13k bots, which were later removed by Twitch)
- 665 maximum number of viewers on stream
- 89 average number of viewers on streams
- 12 909 unique chat users
- 2.35M messages in the chat
- 5 977 user-created clips
- 844 paying subscribers
- $20 640 in total expenses on the channel (including streamers’ paychecks, advertising budget and other activities)
- 4 473 unique Playkey visitors (final CPC is $4,62)
- 495 unique users brought to Playkey (CR from visitors to paying users is 11.06%)
- $41,69 customer acquisition cost
- $3 486 is the income from acquired users (the sum might increase, as some of the users are still gaming with us, but at ARPPU of $7,03 and CAC of $41,69 this simply doesn’t work)
- $7 275 received in donations
- $1 908 received from paying subscribers
We shut down the channel in March. We still had no understanding of how long it would take before becoming profitable. The growth rate was rather slow, yet it required a lot of micromanagement and time from me: planning the schedule, setting up marketing campaigns, creating and planning new shows, interacting with the audience, etc.
Twitch definitely was a great tool for increasing loyalty and managing the community, it became a place of regular meetings for veteran users of the service. It helped deliver news and sometimes share inside information and even a tech support station at times. However, in the rapidly growing market, the main focus for us was on user acquisition.
Are you looking for a platform to communicate with the users? If so, Twitch could work well for you. It is a perfect platform for AMA sessions, getting user feedback, gaining an audience for cust dev or delivering information. Twitch chat works much better than VK or Facebook as a place to talk to customers and get to know them.