The Attention Budget Rule: why tutorials should teach maximum 3 things at a time
two & a half gamers · 2026-05-07
💡 Quick Take
1. Understand player engagement by combining facial analysis and gameplay recordings.
2. Use AI to pinpoint critical moments in gameplay that resonate with players.
3. High ROCA scores (like 89%) indicate a model's accuracy in detecting player engagement.
4. Engagement is a blend of caring (emotional expression) and paying attention (focus on the screen).
5. Normalize emotional and attention data for each individual player, as expression levels vary.
6. Prioritize the first 15-60 minutes of a game for playtesting, as it's crucial for retention.
7. Avoid interrupting combat with tutorial messages, especially in early, easy fights.
8. Teach mechanics sequentially, allowing players to immediately use and see the impact of what they've learned.
9. Limit the number of new mechanics taught at once to a maximum of three to avoid overwhelming players.
10. Repeat tutorial prompts periodically until players demonstrate understanding.
11. Space out tutorial content and allow for "downtime" for players to process information.
12. Avoid chaining multiple passive experiences (victory screens, loading screens, cutscenes) back-to-back.
13. High engagement moments include combos, special moves, art style, team management, and story dialogue in small doses.
14. Low engagement moments include tutorial overload, in-fight combat prompts, end-of-level passive chains, text-heavy loading screens, and button mashing.
15. Aim for dynamic engagement curves, not constant highs, to prevent player burnout.
16. A neutral, active state is good, but dropping into a passive, disengaged state is detrimental.
17. Don't mistake difficulty for lack of engagement; challenging moments can be highly engaging if mechanics are understood.
18. The curve should oscillate between focus and relaxation, mimicking a flow state.
19. For usability testing, a smaller sample (e.g., 7 users) can identify issues, while statistical significance requires 25-30 users.
20. For creative testing, a larger sample (100+ players) is needed for statistical significance due to shorter content exposure.
21. Creatives need initial tension and "threats" to hook players, followed by relaxation and then renewed stress.
22. Failure moments in ads can be highly engaging as they prompt players to consider how they could have done better.
23. Call-to-actions should appear during high-engagement moments in ads.
24. AI can help identify specific elements within creatives that drive performance, not just overall success or failure.
25. Recommendations for creatives should be conversation starters, not definitive answers, as client expertise is crucial.
26. The execution of a concept matters as much as the concept itself in ad creation.
27. Distinguish between "AI ads" that are well-executed and those that are not.
28. Tagging specific moments of engagement within ads allows for iterative improvements and repetition of successful hooks.
29. Player behavior in-game can differ significantly from their self-reported survey responses.
30. Discrepancies between behavior and surveys are opportunities to understand communication failures and player perception.
31. Avoid letting players design games; they may not always know what they truly want or need.
32. Social desirability bias can lead players to disavow engaging content they secretly enjoy (e.g., pimple popping ads).
33. Understand the underlying appeal of successful creative concepts to develop less derivative but equally effective alternatives.
📊 Detailed Explanation
1. Understand player engagement by combining facial analysis and gameplay recordings. This is the foundational methodology. By capturing players' facial expressions and their in-game actions simultaneously, you get a rich, multi-dimensional view of their experience. Facial expressions provide direct emotional cues, while gameplay recordings show what's happening in the game that might be eliciting those reactions. It's like having a direct line into the player's mind and heart.
2. Use AI to pinpoint critical moments in gameplay that resonate with players. Raw gameplay footage is hours long. AI acts as a powerful filter, sifting through all that data to identify the specific segments where players are most invested. This saves developers immense time and resources by focusing their attention on what truly matters to the player experience, rather than trying to analyze everything manually.
3. High ROCA scores (like 89%) indicate a model's accuracy in detecting player engagement. ROCA (Receiver Operating Characteristic Area) is a statistical measure of a model's ability to distinguish between engaged and unengaged players. An 89% score means the AI is correct about whether a player is engaged or not about 89 times out of 100. This is a significant indicator of the AI's reliability and the scientific rigor behind the analysis.
4. Engagement is a blend of caring (emotional expression) and paying attention (focus on the screen). The key insight here is that engagement isn't just one thing. It's a combination. A player might be looking at the screen but not really processing it (low attention), or they might be reacting emotionally to something happening off-screen (high emotion, low attention to game). True engagement requires both: emotional investment and cognitive focus on the game content.
5. Normalize emotional and attention data for each individual player, as expression levels vary. People express themselves differently. Some are naturally very animated, while others are stoic. The AI accounts for this by normalizing the data. What might be a subtle twitch for one player could be a significant emotional response for them, and the system recognizes that by comparing their reactions to their own baseline, not to a universal standard.
6. Prioritize the first 15-60 minutes of a game for playtesting, as it's crucial for retention. This is a critical takeaway for game developers. No matter how amazing the rest of your game is, if players don't connect with the initial experience, they'll never see it. The first hour is the "make or break" period for hooking players and encouraging them to invest more time.
7. Avoid interrupting combat with tutorial messages, especially in early, easy fights. This is a common pitfall. Players are in the zone during combat, enjoying the action. Pausing that flow to explain an optional mechanic they don't even need for an easy fight is jarring and frustrating. It breaks the immersion and can make the game feel less satisfying.
8. Teach mechanics sequentially, allowing players to immediately use and see the impact of what they've learned. Information has no value if it's not immediately actionable. When a player learns a new move or mechanic, they need a chance to try it out and experience its benefit right away. This creates a "aha!" moment and reinforces learning far better than just reading about it.
9. Limit the number of new mechanics taught at once to a maximum of three to avoid overwhelming players. Our brains have a limited capacity for processing new information at any given time. Trying to teach too many things at once leads to information overload, where players remember none of it. Sticking to a small, digestible number of concepts ensures better comprehension and retention.
10. Repeat tutorial prompts periodically until players demonstrate understanding. Players forget things, especially in the heat of gameplay. A single tooltip isn't enough. Implementing a system that re-presents information if a player hasn't used a mechanic in a while ensures that the knowledge sticks and is available when it's actually needed.
11. Space out tutorial content and allow for "downtime" for players to process information. Learning isn't a constant barrage. Players need moments to breathe, digest what they've learned, and integrate it. This means not just teaching things sequentially but also giving them time and space to practice and understand before introducing something new.
12. Avoid chaining multiple passive experiences (victory screens, loading screens, cutscenes) back-to-back. While each of these elements is fine on its own, stringing them together creates a "dead zone" where the player is just waiting. This can be incredibly boring and disengaging, especially after an exciting combat encounter. The combined effect is much worse than the sum of its parts.
13. High engagement moments include combos, special moves, art style, team management, and story dialogue in small doses. These are the elements that truly capture players' attention and create excitement. Flashy animations, strategic depth, visually appealing aesthetics, and well-delivered narrative snippets all contribute to a positive and memorable experience.
14. Low engagement moments include tutorial overload, in-fight combat prompts, end-of-level passive chains, text-heavy loading screens, and button mashing. These are the common friction points that can pull players out of the experience. Overwhelming them with information, interrupting action, boring them with passive sequences, or allowing them to succeed through mindless button mashing all detract from the game's quality and player retention.
15. Aim for dynamic engagement curves, not constant highs, to prevent player burnout. Players don't want to be at peak excitement 100% of the time. That's exhausting and leads to burnout. A good experience has peaks and valleys, allowing players to relax and recover before the next intense moment. This dynamism keeps the experience fresh and sustainable.
16. A neutral, active state is good, but dropping into a passive, disengaged state is detrimental. There's a sweet spot where players are actively involved but not overwhelmed. This is a healthy state. However, when engagement drops below a certain threshold into passive observation or boredom, that's when players start to tune out and potentially leave.
17. Don't mistake difficulty for lack of engagement; challenging moments can be highly engaging if mechanics are understood. Sometimes, analytics might suggest making a game easier because players are struggling. But if players are struggling because they're deeply focused and trying to master mechanics, that's actually a sign of high engagement. The key is ensuring they understand *how* to overcome the challenge, not just making it simpler.
18. The curve should oscillate between focus and relaxation, mimicking a flow state. The ideal player experience isn't a flat line. It's a dynamic back-and-forth between intense concentration and moments of calm. This oscillation, similar to the psychological concept of "flow," keeps players engaged and provides a satisfying rhythm to gameplay.
19. For usability testing, a smaller sample (e.g., 7 users) can identify issues, while statistical significance requires 25-30 users. For quick, qualitative feedback on usability, a small group can reveal major problems. However, to confidently say "X% of players experience this," you need a larger sample size to achieve statistical significance, meaning the results are unlikely to be due to random chance.
20. For creative testing, a larger sample (100+ players) is needed for statistical significance due to shorter content exposure. Ads are typically very short (e.g., 30 seconds). To get reliable data on how players react to these brief exposures, you need to show them to a much larger audience to ensure the results are representative and statistically sound.
21. Creatives need initial tension and "threats" to hook players, followed by relaxation and then renewed stress. Ads need to grab attention immediately. This often involves presenting a problem or a threat within the first few seconds. After this initial hook, there should be a brief period of relaxation before reintroducing tension. This dynamic creates a compelling narrative that keeps viewers watching.
22. Failure moments in ads can be highly engaging as they prompt players to consider how they could have done better. When an ad shows a character failing, it taps into the viewer's desire to "do better." They see themselves in that situation and think, "I could have solved that!" This creates a personal connection and a motivation to download the game to prove they can succeed.
23. Call-to-actions should appear during high-engagement moments in ads. The crucial moment to ask for a download or click is when the viewer is most invested and excited. If the call-to-action appears during a lull or a low-engagement part of the ad, it's likely to be ignored. Timing is everything.
24. AI can help identify specific elements within creatives that drive performance, not just overall success or failure. Instead of just knowing an ad worked or didn't, AI can break down *why*. It can highlight specific visual elements, narrative beats, or gameplay mechanics within the ad that elicited positive or negative emotional responses, allowing for more targeted improvements.
25. Recommendations for creatives should be conversation starters, not definitive answers, as client expertise is crucial. The AI provides data-driven insights, but the people who make the ads have invaluable experience and intuition. The AI's suggestions should be a jumping-off point for discussion, allowing clients to refine or reject ideas based on their deep understanding of the market and their product.
26. The execution of a concept matters as much as the concept itself in ad creation. Having a great idea for an ad is only half the battle. How that idea is brought to life visually, aurally, and narratively makes a huge difference. A brilliant concept can fall flat with poor execution, while a simpler concept can shine with masterful delivery.
27. Distinguish between "AI ads" that are well-executed and those that are not. The term "AI ads" can be generic. There's a significant difference between ads that leverage AI effectively for compelling storytelling and those that are simply basic concepts presented poorly. The quality of execution is paramount, regardless of the underlying technology.
28. Tagging specific moments of engagement within ads allows for iterative improvements and repetition of successful hooks. By identifying the precise moments in an ad that grab attention, developers can learn what works. They can then create variations that amplify these successful hooks or apply them to different concepts, leading to a more efficient and effective creative process.
29. Player behavior in-game can differ significantly from their self-reported survey responses. What players *say* they like or dislike in a survey often doesn't align with their actual actions and emotions while playing. This is a common phenomenon; people may rationalize their behavior or not fully recall their in-the-moment feelings.
30. Discrepancies between behavior and surveys are opportunities to understand communication failures and player perception. These differences aren't about one signal being "right" and the other "wrong." They are valuable insights into how the game is being perceived versus how it's intended. They highlight areas where communication could be clearer or where the player experience might be unintentionally misaligned with developer goals.
31. Avoid letting players design games; they may not always know what they truly want or need. While player feedback is vital, directly translating player requests into game design can be problematic. Players often express immediate desires without understanding the broader implications or their own underlying motivations. Developers need to interpret feedback through the lens of overall game design principles.
32. Social desirability bias can lead players to disavow engaging content they secretly enjoy (e.g., pimple popping ads). People are influenced by what they perceive as socially acceptable. Even if a player finds a certain type of content engaging (like a gross-out ad), they might deny enjoying it in a survey because it's not something they want to admit to publicly. This is a significant factor in understanding survey data.
33. Understand the underlying appeal of successful creative concepts to develop less derivative but equally effective alternatives. When certain ad concepts are highly successful (like "save the character" or "pimple popping"), the goal isn't just to copy them. It's to understand *why* they work – what core human desires or psychological triggers they tap into – so you can create new, original concepts that achieve the same effect without being embarrassing or overused.
🎯 Expert Opinion
This discussion really highlights the evolution of game testing and ad creative analysis, moving beyond simple metrics to a deeper understanding of player psychology. The integration of AI for facial and gameplay analysis is a game-changer, providing objective, real-time data that complements subjective player feedback. The emphasis on the "moments that matter" is crucial – developers can't afford to waste player attention on anything less.
The insights into tutorial design are particularly valuable. The concept of "attention budget" and the need for sequential, actionable learning are fundamental. This isn't just about fighting games; it applies to any genre with complex mechanics. The idea of repeating prompts until mastery, rather than expecting instant retention, is a lesson learned from decades of game design, and it's great to see it reinforced with data. The critique of passive end-of-level sequences is spot-on. We've all experienced those moments where the game takes control for too long, breaking the player's immersion and momentum. This is where careful pacing and player agency become paramount.
For ad creatives, the focus on dynamic tension and release is a masterclass. The "near-death experience" concept, while perhaps a bit extreme in its phrasing, captures the essence of creating compelling hooks. Ads that present a problem and then hint at a solution, or show a character in peril, tap into primal human instincts for problem-solving and survival. The fact that failure moments can be more engaging than success highlights the power of psychological reactance – players want to prove they can overcome challenges. This is a powerful insight for anyone creating marketing materials.
The discussion on the discrepancy between survey data and actual behavior is a classic challenge in user research. The pimple-popping ad example is perfect – it illustrates how social desirability can skew self-reported data. This is why behavioral data, like that captured by Mhance's AI, is so critical. It provides a more honest, unvarnished view of player reactions. The challenge for creatives and developers is to find the "socially acceptable" equivalent of these engaging, albeit sometimes "embarrassing," concepts. This is where true innovation lies – understanding the core appeal and translating it into universally appealing experiences.
Looking ahead, the trend towards AI-driven, granular analysis of player experience is only going to accelerate. We'll see more sophisticated models that can predict player churn, identify optimal difficulty curves, and even suggest narrative adjustments based on emotional responses. The ability to dissect creatives at such a fine level, identifying not just what works but *why* it works, will revolutionize UA strategies. It moves us from broad A/B testing to highly targeted, data-informed creative iteration. The future is about understanding the player's emotional journey, moment by moment, and using that insight to build more engaging games and more effective advertising.
Kanal: two & a half gamers