Why game studios keep solving the wrong problem
Player Behaviour, Algorithms, Emotion, and Studio Game Decision-Making
Justin French - CEO & System Architect - GameDataCore · Jun 23, 2026
Game development teams are constantly asked to make hard decisions with incomplete information. Who is this game really for? Why will players care? Is the problem the trailer, the onboarding, the price, the fantasy, or simply that the team is out of budget?
My almost 20 years working in games have shown me that most studio decisions happen amid chaos, and too often that chaos produces the loudest-voice-in-the-room syndrome: uncertainty filled by hierarchy, instinct, or whoever sounds most certain
A game is not merely a creative or artistic endeavour, and a launch is not merely a commercial event. It is an interaction between players, teams, platforms, emotions, money, and probability. None of these parts can be understood in isolation. To make better decisions, studios need to understand four interacting systems: the platform system that shapes discovery, the psychological system that shapes player meaning, the production system that shapes what teams can actually change, and the commercial system that shapes risk. Better decisions come not from mastering one lens, but from holding all four at once — and then asking what caused the outcome you observed.
The platform system: visibility as a feedback loop
The platform system operates as an adaptive feedback loop. Steam, the dominant PC discovery platform, estimates the likelihood a user will engage with a game based on behavioural signals — clicks, wishlists, purchases, play sessions, reviews (Cheuque, Guzmán & Parra, 2019). In Bayesian terms, the system starts with a prior belief about where a game belongs, then updates that belief as evidence arrives. Valve’s own documentation alludes that visibility is driven by player interest rather than paid placement or any single metric (Peterson, E. / Valve). The algorithm does not understand creativity; it only sees behavioural traces.
The causal question here is not “are our numbers going up?” but “which signals are driving visibility, and which are merely correlated with it?” A wishlist spike after a festival appearance may reflect genuine audience fit, or it may reflect curiosity that never converts. Treating the spike as evidence of product-market fit without asking what caused it is one of the most common and costly mistakes a team can make.
The psychological system: meaning behind the signal
Platform behaviour is not the same as player meaning. A click signals curiosity. A wishlist signals expectation. A purchase signals intent. Playtime may reflect satisfaction, habit, social need, or any other motivational factor. A refund might signal disappointment — or that the game reached the wrong player entirely.
Self-Determination Theory holds that intrinsic motivation depends on satisfying three basic needs: competence, autonomy, and relatedness (Deci & Ryan, 2000). Two players might both leave negative feedback about combat, but one is frustrated (competence blocked by unclear controls) while the other is bored (competence and autonomy undermined by shallow mechanics). Without a psychological framework, both collapse into “negative sentiment,” and the team risks solving the wrong problem entirely.
The causal lens matters here, too. If changing the onboarding improves retention, did the onboarding cause it — or did a concurrent community post, a streamer pickup, or a platform recommendation explain the same effect? Attributing the outcome to the wrong cause leaves the underlying problem unaddressed and creates false confidence in a fix that may not hold.
The production system: what can actually be changed
Even when a team correctly interprets evidence, the production system determines what is actionable. Budgets, deadlines, technical debt, and publisher expectations constrain which problems can be addressed. A studio might know that onboarding needs additional work, but lacks the runway to fix it. It might discover the game resonates with a sharper audience while stakeholders push for broader appeal.
These are not analytic failures — they are structural tensions inherent to creative production. The value of holding the production lens is not to resolve those tensions but to be honest about them. A team that knows it cannot fix onboarding can at least stop debating whether to fix it, and redirect energy toward what is within reach: a clearer content warning, a better store page, a more targeted trailer cut. Clarity about constraints is itself a decision input.
The commercial system: the boundaries of tolerable risk
The commercial system sets the boundaries within which all other decisions operate. Development budgets, platform revenue shares, and investor expectations define tolerable risk. A decision that makes creative and psychological sense — delaying launch to refine onboarding — may be commercially impossible. Conversely, cutting scope to ship on time may damage the player experience, eroding long-term revenue and player trust.
What the commercial lens adds beyond “we have limited money” is a framework for prioritising under pressure. When a studio knows which decisions affect the metrics that investors or publishers are actually watching — wishlist conversion, peak concurrent users, review score at launch — it can make scope and timing decisions that are commercially legible, not just creatively defensible. The team that can say “we cut this feature because it would not have moved the number that matters most” is better positioned than one that cut it because they ran out of time.
Why the lenses have to work together
The value of an interdisciplinary approach lies in how these systems interact. A drop in wishlist conversion could indicate poor trailer messaging (production), a mismatch between marketing fantasy and game experience (psychological), exhausted ad spend (commercial), or a shift in platform visibility (platform). The same surface signal means something different depending on which lenses are applied together — and which causal question is being asked.
The exploration-exploitation trade-off from reinforcement learning captures this tension: studios must balance learning from uncertain directions with exploiting known successes (Sutton & Barto, 2018). Confounders are everywhere — a concurrent influencer campaign, a competitor’s patch, seasonal playtime shifts, changing platform recommendation weights. Any one of them could independently explain an observed outcome, and the attempt to isolate a single cause can produce false confidence as readily as genuine insight.
Better decision-making in games is therefore not about eliminating uncertainty. It is about reasoning more rigorously within it: naming which system you think is driving an outcome, stating what evidence would change your view, and tracking whether the decisions you made actually produced the results you predicted. That last step — closing the loop — is the one most studios skip.
What this means in practice
Until studios, publishers, and investors treat game development as a system of interacting perspectives, they will keep making bets that fail for reasons they should have recognised earlier. The antidote is not more data. It is a structured way to connect evidence to decisions, track outcomes over time, and build the kind of institutional reasoning that survives a leadership change, a publisher review, or a post-mortem.
This is the problem GameDataCore is built to address — linking player feedback, behavioural signals, market context, and decisions into a single evidence layer that teams can use to make and defend the calls that matter. If this framework resonates with how your studio thinks, or how it should think, we’d like to show you how it works in practice.
References
Cheuque, P., Guzmán, J. & Parra, D. (2019). Recommender systems for online video game platforms: The case of Steam. Proceedings of the 13th ACM Conference on Recommender Systems.
Deci, E. L. & Ryan, R. M. (2000). The “what” and “why” of goal pursuits: Human needs and the self-determination of behaviour. Psychological Inquiry, 11(4), 227–268.
Peterson, E. / Valve. Steam visibility: How games get surfaced to players. Steamworks presentation.
Sutton, R. S. & Barto, A. G. (2018). Reinforcement learning: An introduction. MIT Press.
Pearl, J. & Mackenzie, D. (2018). The book of why: The new science of cause and effect. Basic Books.