“Juice” is one of the most useful words game developers have and nobody else uses. It names the layer of rich interaction-contingent feedback that makes an action feel like it happened — the squash on a bounce, the particles on a hit, the sound, the slight screen shake, the number that pops up and arcs away.
The canonical demonstration is Martin Jonasson and Petri Purho’s 2012 talk Juice it or lose it, in which the same Breakout clone goes from lifeless to delightful with no change to the rules whatsoever. Everything added was feedback.
Juicy Interactive Visualization: Evaluating How Excessive Feedback Design Shapes Visualization Engagement, posted 6 October 2026 by Shano Liang, Max Chen and Lane Harrison, asks whether data visualisation should borrow it.
The study
JuicyVIS, a framework translating juicy feedback concepts into visualisation, with three dimensions:
- Interaction type
- Feedback timing
- Feedback intensity
Explored through 26 prototypes, across three mixed-methods online experiments.
The findings
Juicy feedback consistently correlated with increased engagement and aesthetic satisfaction.
And the result that makes the paper worth reading:
Post-interaction feedback proved most valuable, not simply intensifying feedback volume.
Timing beat intensity. Making the feedback bigger, louder or more abundant was less effective than placing it after the interaction completed.
Why that is the interesting answer
It is not the obvious one. The intuitive model of juice is more — more particles, more sound, more motion — and the paper’s own title frames the question as being about excessive feedback.
What they found instead points at something about what feedback is for. Three readings, all consistent with the result:
Confirmation needs to arrive after the act. Feedback during an interaction competes with the interaction — you are still dragging the brush, and animating the chart while you drag adds motion you have to see past. Feedback after it completes is confirmation, and confirmation is what tells you the system received your intent.
It respects the task. Visualisation is instrumental in a way a game is not. A player’s goal is the interaction; a visualisation reader’s goal is the data behind it. Feedback that obstructs reading is a cost, not a flourish, which is why “turn it up” has a ceiling here that it does not have in games.
And post-hoc feedback can carry information. A transition that animates to the new state after a filter is applied shows you what changed — which elements left, which remained, where the new values are relative to the old. That is not decoration; it is the chart explaining its own update. Heer and Robertson’s work on animated transitions in statistical graphics established this years ago, and this result is consistent with it.
What to do with it
For anyone building interactive visuals — a dashboard, a data piece, an audio-reactive visualisation, a generative interface:
Add feedback at the end of an interaction, not throughout. Let the drag be quiet and the release be expressive.
Animate the transition, not the input. A 300–500ms eased transition to the new state is the highest-value juice available in visualisation, and it is one line in D3 or an easing curve in whatever you are using.
And don’t mistake intensity for quality. The study’s own framing invites it, and the answer came back the other way.
The broader point: the game-feel literature is a genuinely under-mined resource for creative coding. Juice, game feel, coyote time, input buffering — these are decades of carefully developed craft knowledge about making interaction feel good, written down by people who tested it on millions of players, and almost none of it has crossed into visualisation, installation or tool design. This is the first study we have seen that treats it as a transferable body of work rather than a genre convention.
The same three authors also posted reVISit-XR the same day, which is about the infrastructure for running studies like this one in XR.