As a game developer, whether you aim to create immersive gameplay, improve key product metrics, or maximize revenue, player satisfaction is essential for success. When players have fun and enjoy your game, it drives engagement and benefits everyone involved.
Today’s mobile game market is oversaturated, with every successful idea quickly copied, making it challenging to stand out. Achieving top success is no longer about luck; it’s the outcome of focused effort, thorough research, and data-driven strategies.
Many companies rely exclusively on product analytics, which only measure a feature’s performance after it’s launched. This approach to testing ideas can be slow and expensive. While analytics tools show how game metrics change after an update, they don’t reveal why users respond positively or negatively to a feature or provide insights on how to enhance it.
With this in mind, we developed a method to test any game content before implementation. This approach combines neuroscience with behavioral and emotional analysis, turning these insights into metrics that align closely with player engagement.
We adapted our emotion recognition methodology, originally designed for retail, to suit the unique demands of mobile game development. With these insights, developers can accurately analyze and refine the player experience to maintain high engagement. Here’s how we make it happen.
Understanding core metrics of emotional engagement
What does “fun” really mean?
When we analyzed an Overwatch-like mobile shooter, one of our respondents had their character hit by a train during a game session, resulting in a loss of match. Technically, this could be seen as a failure, right? However, the player was so excited by this unexpected interaction with the level design that their arousal spiked. Rather than feeling discouraged, they became more motivated to overcome the loss. Even though the situation was negative, what matters the most is that arousal increased. It shows that something was out there. This approach is used in Souls-like games, where players may feel frustration from challenges, but what matters most is that they are emotionally engaged.
To visualize emotional engagement, there is one key metric: arousal. Many of you can giggle at the name of that metric because it’s often being used in different contexts, but for scientists and researchers “arousal” is a valid and widely recognised term.
Arousal shows the level of emotional engagement of the audience during interaction with content, regardless of reaction polarity (positive/negative). Players' arousal is vital for immersion, gauging how effectively it captivates and sustains players' attention. And it directly correlates with retention.
Every game designer is familiar with this chart. It illustrates the balance between a player’s skill level and the challenge the game presents, representing the "ideal" flow of player progression. Now, compare it with the chart below, which shows what a successful game might look like based on our tests. While arousal can and should fluctuate, the overall trend needs to remain positive.
How it works?
We use the combination of three neural networks and the brainpower of human researchers to make it work. The first neural network assesses an individual’s emotional baseline and monitors changes in trackable emotions. It uses a neurophysiological scientific framework to transform emotional responses into arousal. Another neural network focuses on tracking gaze direction, while the third one aggregates second-by-second data into a report that humans can read.
Paul Ekman’s emotions classification serves as the basis for our arousal model development. We track seven emotions and convert them into one simple yet indicative metric: arousal.
As previously mentioned, the polarity of emotions is not a concern for us. Whether we manage to frighten or delight the player, the key is that we've captured their attention and kept them engaged. Every emotional response is valuable as long as it's not caused by a critical interface bug that makes the player lose progress.
Tracking these emotional reactions provides us with clear and objective data about players' perceptions, which we wouldn’t be able to obtain through participant answers or surveys alone.
It looks like we've successfully captured shifts in players' emotional responses. But how do we ensure the data we collect is reliable? The way we structure research plays a key role. We don't set specific goals but instead let players explore the game freely, measuring how they feel about your product. Our analysts then cross-check survey data for any meaningful correlations. Unlike traditional playtests, we don't heavily rely on questionnaires, as responses can be influenced by what participants think we want to hear. Emotions, however, don’t lie. In our methodology, we filter out emotional fraud - such as blinks or talking moments - to ensure the emotional response data is clean and reliable.
Hack game design
When we analyzed the first play session (FTUE) of an action shooter game, the test revealed an issue with the timing of introducing the battle pass. It was presented after the player had already received their first reward, which caused confusion. The player did not understand how the battle pass system worked, as its introduction did not coincide with the reward's context or explanation. While these tweaks might seem obvious, they can easily be overlooked due to a simple but crucial fact: games are made by people, and making games is hard.
Emotion tracking opens up many new possibilities. It allows us to understand why tutorial conversions fail, which competitor mechanics are more appealing, and which elements in the first play session engage users to stay. With these insights, we provide the opportunity to consciously shape the player experience.
Want to learn more about how Sensemitter can improve your game? Contact us via this link.