Find the core action
Describe the game with one verb: place, dodge, connect or choose. Write down what the player sees, the decision they make and the feedback they receive. If the explanation needs a long feature list, narrow the idea.
For a tower game, the loop might be waiting, releasing a block and judging its landing. A shop or map does not yet help test that moment.
Build one experiment
Use simple shapes and one level. Include input, success, failure and a quick restart. Add clear visual or sound feedback so the action has a readable consequence.
Change one parameter at a time, such as movement speed or the landing zone. Otherwise you will not know which adjustment changed the experience.
Observe without guiding
Let someone play without instructions wherever the interface should explain itself. Record hesitation, misunderstood intentions and unclear outcomes. Ask about specific moments.
A voluntary retry is a useful signal, but one successful session does not validate the whole game. Test with other people and distinguish understandable controls from an engaging mechanic.
Make a decision
Finish with a short decision: keep the mechanic, change one part or discard the variant. Save observations and prototype settings to avoid repeating the same experiment.
Add progression, content and a visual system after this decision. A prototype saves work when it selects a direction rather than reproducing the future game at a smaller scale.