"Being open to AI solutions is extremely important, otherwise you're missing out"
"Being open to AI solutions is extremely important, otherwise you're missing out"
Gamelight's Günay Aliyeva talks automated user acquisition, dynamic rewards, and the transformation of mobile game marketing.
Welcome to the latest Q&A edition of AI Gamechangers, your newsletter where AI meets games. Each week, we’ll send you an interview with a leading industry figure. Remember, you can read them all in the archive.
This week, we’re speaking with Günay Aliyeva, the co-founder of Gamelight, the rewarded marketing platform for mobile games. Gamelight's AI algorithm analyses users' playtime, engagement, competitor game usage, and demographic data to identify users with the highest likelihood of long-term engagement.
It’s a big week for AI news. Scroll to the end to read about Anthropic’s updated model and its “computer use” capability, 49 games companies signing up to SAG-AFTRA’s interim agreements, NCSOFT spinning off four independent games companies with one focusing on AI, and more.
Be sure to check back next week when we’ll dive into emergent gameplay in a new AI-powered sim called AuraVale. And we’ve got tons of other interviews in the bag, too, so stay tuned.
Günay Aliyeva, Gamelight
For your last AI|G encounter of October, meet Günay Aliyeva, co-founder of Gamelight. Established in 2022, Gamelight was recently listed as the best rewarded UA source globally on the AppsFlyer Performance Index. We talk about the move away from traditional manual targeting to more sophisticated behavioural matching, the drive to simplify campaign management, and overcoming initial scepticism about AI optimisation by demonstrating strong results.
Top takeaways in this interview:
Embracing AI is essential. Rather than replacing humans, it’s a tool that can take on data-driven tasks, leaving people free to think creatively and strategically. Eventually, AI adoption will be a competitive necessity rather than an option.
Gamelight has developed an AI-driven platform that automates and optimises mobile game user acquisition, replacing traditional manual targeting methods. Rather than using fixed reward structures, the platform adjusts the ratio of reward types based on how each user responds.
Simplification and speed is key. The set-up process takes minutes, significantly reducing the time and effort needed to start campaigns.
AI Gamechangers: Please give us some insight into your background. How did Gamelight come together?
Günay Aliyeva: I’ve been in this industry for more than 10 years. By this industry, I mean mobile apps and mobile gaming. I’ve worked at different mobile companies and tech companies, both on the demand and the supply side, as a service provider and as a client. I have been founding and scaling my own companies for the past five years.
We’ve been doing a lot of user acquisition within mobile apps, and I have started noticing how unoptimised everything is. It includes a lot of manual, routine work. Launching is a lot of effort, from figuring out the dashboard (if there is one) to trying to find the correct person if it’s a managed account. After launching, it’s a lot of work to optimise campaigns because there’s no straightforward way to do it. You need to go through numerous data points to see if they make sense. In many cases, it's nearly impossible to do this effectively — after all, how much data can a person realistically analyse?
There are a lot of things that could be better — so we decided to launch Gamelight. First of all, we have our own self-serve dashboard, which is not common within the rewarded channels (these features are typically found in video networks or social platforms). We made it very easy to launch campaigns. With other dashboards, for example, it could easily take you an hour to figure things out once you register. Uploading creatives, creating campaigns, ad groups, setting up the integration, and so on, often take a lot of time.
With our platform, however, it only takes a few minutes to get everything up and running. In just two to three minutes your campaign is ready to launch, all you need for setup is a tracking URL. We don’t require creatives because we fetch everything directly from the store. Any changes or updates in the store are instantly reflected in our apps in real-time, too.
After the launch, we built an AI algorithm that optimised many data points. This marked the point at which AI became deeply integrated into our processes.
In other rewarded channels, rewards are often distributed manually. Perhaps elsewhere, you choose to give a certain number of coins to complete an event or for a certain amount of playtime. We take a different approach based on user behaviour. We see how each user behaves using our apps and games, and how they react to different reward structures, and we give them rewards based on this.
The second aspect is targeting. Many other channels rely on manual targeting — if they implement any targeting at all. When they do, it's often limited to age and gender targeting or OS version targeting. You define some average numbers based on what worked elsewhere, but you probably miss some users that would be very good for you, because they don’t fit into this average group. Let’s say you’re looking for casual gamers, female, 30-plus, right? If a good user doesn’t fit right there, you will probably not target them any more because you’re doing manual targeting. In our case, this doesn’t happen. We do everything based on the AI algorithm that will find the best user without generalising the audiences.
Can you give us a flavour of how Gamelight’s AI algorithm works in this case? How does it analyse user data and what does it consider when recommending games?
When users register on our platform, they share their age and gender and give us access to device data and OS version. Additionally, they need to grant access to app usage data, which means we can see which games they play and which other apps they use, when and for how long.
We analyse those to find patterns between users and see what would be a good fit for the games that we promote. So, for example, if we notice a user who frequently plays casual games, stays engaged for months and owns a high-end device — we can recommend those users to the right game publishers that we think would be the best fit.
Some partners optimise towards the highest engagement and in-app purchases, while others optimise towards the highest engagement and retention. An ad-monetised game, for instance, usually has shorter ROAS cycles. The users would behave differently; they’ll be users who are more engaged in the short term but maybe have a lower attention span, and would try a new game after some months. However, this can still be beneficial for ad-monetised games, as these users can generate significant profit within that timeframe and help ad monetised games reach complete profitability.
But if it’s an IAP monetised game, you want a user that sticks around for at least a year and actively engages with the game to discover all possibilities within it.
Based on this data, we can match users to games and games to users, and it works both ways.
What’s your philosophy? What makes you different to other companies working in user acquisition and monetisation?
Our purpose is to make things easier and better.
Unfortunately, there are a lot of things in this industry that are very unoptimised. It shouldn’t be this way! Each time I talk to someone outside this industry, they ask, “Why do you need that many salespeople?” or “Why are there people in this industry talking and selling things to each other? Why don’t those campaigns work based on what is a fit? Shouldn’t it be based on numbers?” It’s partially true. We need humans for the creative approach, to build strategies, to define the direction, and to come up with ideas. But when there’s data, you don’t need people sitting around updating Excel sheets to analyse it; you can optimise it all with algorithms.
We want to remove the parts that are unnecessary for humans to do. As a result, people can focus on more strategic decisions within user acquisition.
“I see people asking if AI is a good thing. Yes, it’s a good thing! It creates more jobs. You can use it to optimise your work. AI is here. Using AI tools correctly is a huge power.”
Günay Aliyeva
We can also simplify the whole process a bit because it’s very unnecessarily complicated now. For example, launching a UA channel can sometimes take months of back-and-forth communication when it should be straightforward. We know the game we’re promoting, so here’s a tracking link, so let’s start promoting, right? But it can be overly complicated, even for a simple campaign. So, I would say our goal is to simplify the process and make it more optimised for both sides, advertisers and publishers.
We published a few case studies with our partners on our website. They are primarily for two directions: ad-monetised games and IAP-monetised games. In each case, we go into detail about the campaign setup, goals, and how we reached them. Joycity, one of the largest Korean game publishers, is an example, as are Superplay, Stillfront Group, and more.
Other recent standout cases include:
257% D7-D30 ROAS growth for World War: Machine Conquest by JOYCITY
284% D30 ROAS over-delivery for Sunshine Island by Stillfront Group
D30 ROAS over-delivery of 162% for Bloodline: Heroes of Lithas by GOAT Games
D30 ROAS over-delivery of 91% for Legend of Slime by AppQuantum
480% volumes with ROAS D7 at 142% for Word Farm Adventure by Mad Brain Games
Do you think the games industry is receptive to AI-powered marketing solutions? Or have you noticed any resistance and scepticism?
Initially, people were sceptical when we approached them and said, “Hey, you don’t really need to optimise your campaigns; you just launch, and the algorithm will self-learn and self-optimise! You’ll just need to keep an eye on it and decide which direction you want to go.” They just need to define their caps or decide to launch in new geos or something else new. But they don’t need to optimise their campaigns, per se, like they used to.
They were hesitant, saying, “Can we still do manual targeting? Can we still target just women?” But there’s no need anymore! The algorithm will see what’s happening. Not every female user will like your game, and not every male user will dislike your game.
At first, explaining how it worked was challenging. But once word spread and people saw the campaigns in action, our advertisers began reaching out to their industry friends and recommending us, explaining that it works very well. When others hear about its effectiveness, they start trusting the product more. While some partners may still prefer manually setting targets or rewards, 99% are very open to this better solution.
What advice do you have for people in the games industry considering AI solutions? What should they keep in mind?
First of all, they should be open to these solutions. Unfortunately, way too often on LinkedIn, I see people asking, “What is AI doing to our jobs?!” They ask if AI is a good thing. Yes, it’s a good thing! It creates more jobs. You can use it to optimise your work. It’s not about “AI came along, and now I have nothing to do.” AI is here; you have something else to do — use the tool.
“Our purpose is to make things easier and better. When there’s data, you don’t need people sitting around updating Excel sheets. We want to remove the parts that are unnecessary for humans to do. People can focus on more strategic decisions.”
Günay Aliyeva
Using AI tools correctly is a huge power in my opinion. Many people still don’t know how to handle them, and learning to use them could greatly benefit the market. We saw how companies that weren’t open to those solutions missed out compared to those that embraced them quickly, launching and scaling the campaigns at an impressive pace.
Being open and actually learning about those tools is extremely important. Try out as many as possible. We are using a lot of AI tools ourselves. And if there’s something new, it’s worth testing it even for the short term and knowing for yourself if it’s good or not, making an informed decision. Otherwise, you’re potentially missing out on something that could greatly impact your business.
Further down the rabbit hole
It’s been a big week for AI news. Here are some useful links. Enjoy!
This week, Anthropic unveiled its upgraded Claude 3.5 Sonnet model and the new Claude 3.5 Haiku. As well as improved coding abilities, Anthropic hit the news with its “computer use” capability - yep, Claude is the first frontier AI model in public beta that can move your cursor, click links and type text on your device.
It has emerged that TikTok’s parent company, ByteDance, fired an intern this summer for deliberately sabotaging the training of its AI model.
Last week, the International Federation of Actors (FIA), representing performers from some 60 countries, unanimously voted to support SAG-AFTRA's ongoing video game strike, particularly backing their fight against AI exploitation. This week, SAG-AFTRA revealed that over 120 games from 49 companies have signed its interim agreements.
Top Korean games company NCSOFT will create four independent companies, one focusing on AI. The new AI company will work on NCSOFT’s self-developed LLM Varco. If approved next month, the spinoff will happen in February.
Unity's Supersonic has launched an AI-powered Game Idea Generator that helps developers create and refine game concepts by providing descriptions, mechanics, monetization strategies, and initial code for Unity engine development.
The inaugural AI and Games Conference takes place on Friday 8th November in London, and the final speakers have been announced. They include Yassine Tahi (Kinetix) and Roberto Lopez Mendez (ARM).
Yassine Tahi, who will speak at the AI and Games Conference, was our interviewee on September 23rd. Kinetix, the 3D emote company, won the Best AI Games Tech category in this year’s Mobile Games Awards.
