A HUD can look busy and still tell an experienced player exactly what they need. It can also look beautifully minimal while leaving a new player unsure what just happened. A screenshot alone is a weak basis for deciding what to remove.
View full image
Start with an interaction. What did the player intend to do, what did the system consider eligible, which input did the player use and what happened next? That sequence exposes questions a layout review cannot answer.
Follow the action all the way through
For a contextual interaction, examine the moment before the button press. Can the player identify the potential target? After focusing, can they tell which commands are available? After acting, is the result visible? If the action was unintended, can they recover?
These are related but separate problems. A readable “Talk” prompt does not guarantee that the player knows whom they will address. A clear selection marker does not explain why an action is disabled.
In my RDR2 study, I explored target markers and prompts because command visibility alone did not resolve the recipient question. One alternative used a portrait in the HUD, but it required comparing a small representation with characters in the scene. A marker closer to the character addressed that comparison more directly.
Every extra cue has a cost
Placing information near a target can shorten that visual comparison, but creates other questions. What happens when characters overlap? Does the cue cover a face? Can it remain legible against movement and changing backgrounds? Is the player watching the world or chasing markers?
A static proposal can suggest an answer. It cannot demonstrate stable behavior across all those conditions. The next prototype should reproduce whichever condition is most likely to break the idea.
Test understanding before asking for preference
Ask a player to identify the target, perform an action and explain the result. Observe hesitation and unintended inputs before asking which version they prefer. Preference can help assess comfort or visual intrusion, but it should not stand in for successful use.
My final controller test involved participants already familiar with the proposed scene. That was useful for evaluating the interaction, and it leaves first exposure as an open question. Naming that limit makes the next step clearer: test unfamiliar players with moving, overlapping targets before calling the approach ready for a real game.