Why Juice Sells: A Frame-by-Frame Teardown of Game Feel

Juice is feedback layered onto a single interaction so the player feels the result before they think about it. It is built from anticipation, impact, and follow-through, timed in frames rather than seconds. Sound carries roughly half of it. Done well it changes nothing mechanical and everything perceptual.

Why this matters

Two games with identical mechanics can test completely differently. The one that feels better holds attention longer and converts better in ad creative, because the satisfying moment is visible in three seconds of footage.

Feel is also cheap relative to almost everything else. It is a few days of work on an existing build rather than a new system.

The trap is treating it as polish to be added at the end. Feel is a design property of the core interaction, and the end is the worst time to discover the interaction has nothing to amplify.

What juice actually is, mechanically

Strip the word back and juice is a set of responses fired by one input.

A single tap that destroys a block might trigger: a scale punch on the block, a particle burst, a screen shake, a brief pause, a sound, a score counter tick, and a colour flash. The block was going to be destroyed either way. Everything else exists to tell the player it happened.

Good feel does not add information the player needed. It confirms information the player already expected, faster than they can consciously read it.

That is why it works on a phone, where the screen is small and the player is often distracted.

Anticipation, impact, follow-through

Animation borrows this structure from traditional practice and it holds up completely in games.

1. Anticipation is the wind-up. A small movement in the opposite direction before the action. It tells the player something is about to happen and makes the impact read as bigger.
2. Impact is the frame of contact. This is where the biggest, sharpest changes go: the scale spike, the flash, the particle spawn, the sound transient.
3. Follow-through is the settle. The overshoot and return that makes the object feel like it has mass rather than snapping to a value.

The common mistake is spending all the budget on impact. An impact with no anticipation reads as abrupt, and one with no follow-through reads as cheap.

Keep anticipation short. On mobile, a long wind-up feels like input lag.

Screen shake, hit stop, and the three-frame rule

These three are the highest-return effects available, and the easiest to overdo.

• Screen shake should be small, fast, and decaying. Directional shake, pushed away from the impact point, reads better than random jitter.
• Hit stop is a very brief freeze at the moment of contact. It is the single most effective way to make an impact feel heavy, and it costs nothing to implement.
• The three-frame rule is a useful default: most micro-effects should resolve inside roughly three frames at 60, which is about 50 milliseconds. Long enough to perceive, short enough that it never delays the next input.

Anything that blocks input for longer than that stops being feel and starts being a cutscene.

A practical starting point for a tap-to-destroy interaction:

Tune by feel on a real device. These are starting values, not targets.

Sound as half of feel

Mute a well-tuned game and roughly half the satisfaction disappears. Players attribute it to the visuals, which is why audio is chronically underfunded.

Three things matter more than sample quality.

• Transient sharpness. A crisp attack reads as impact. A soft attack reads as mush, whatever the rest of the sound does.
• Pitch variation. Randomising pitch slightly on repeated sounds stops the ear noticing repetition. Without it, a rapid-fire action becomes irritating within a minute.
• Layering. A satisfying hit is usually two or three sounds stacked: a click for the transient, a body for the weight, a tail for the space.

Mix for a phone speaker rather than headphones, which we go into in Audio on a Budget.

Particles and the cost of overdoing it

Particles are where feel work meets the performance budget, and they lose more frames than anything else in a casual game.

Keep the count low and the lifetime short. A twelve-particle burst that resolves in a third of a second usually reads better than sixty particles hanging around, and costs a fraction of the fill rate.

Watch overdraw specifically. Large soft particles stacked over each other are the classic cause of a game that runs fine on a flagship and crawls on a budget phone. The fixes are in Unity Mobile Performance.

One interaction, before and after

Take a plain version of a tap that clears a tile. The tile disappears, the score goes up.

Now layer it, in the order we would actually build it.

1. Scale the tile down with a slight overshoot rather than removing it instantly.
2. Add the sound, with pitch variation.
3. Add hit stop of two frames.
4. Add a small particle burst in the tile's colour.
5. Add a directional screen shake, decaying fast.
6. Make the score counter tick up rather than jump, and punch its scale on arrival.

Nothing mechanical changed. The tile still clears and the score still rises. The interaction now reads as an event.

Build them in that order and stop when it feels right. Most interactions need four of the six, not all six.

What we would do

Pick the single interaction the player performs most and spend two days on it before touching anything else. In a casual game that is usually one tap or one drag, repeated thousands of times.

Tune on a mid-tier phone with the sound on, at arm's length, in a lit room. Feel decisions made on a desktop monitor with headphones consistently come out too subtle.

Then set a rule for the rest of the project: no new interaction ships without anticipation, impact, follow-through, and a sound. It is faster to make that the default than to run a polish pass later.

The short version

• Juice is layered feedback on one input, timed in frames.
• Anticipation, impact, and follow-through is the structure worth memorising.
• Hit stop is the cheapest large win available, and screen shake should be small and fast.
• Most micro-effects should resolve in roughly three frames at 60.
• Sound carries about half the perceived quality, so mix for phone speakers.
• Tune the most repeated interaction first, on a real device.

Take the one interaction your player repeats most and layer it this week. If you would like a feel pass on an existing build, tell us what you are working on.

Related reading: Making a 2D Game Feel Good, Input Handling for Mobile, and Why Simple Games Win.