Distance From the Tree

How the study worked

Two boards, twelve images each. Photographs, labels, illustrations and packaging, mixed together with no grouping and no order. Fifty people got a handful of green dots and a handful of red ones, and a single prompt: green on the images you most associate with highest quality orange juice, red on the ones you least associate with it.

They placed 535 dots, and the heatmaps above are the raw result, before any sorting. Both boards ran the identical prompt, which means the same pattern had to appear twice, independently, to count. It did.

At the platform's 60% threshold those 24 images fall into three groups. Eleven cleared the bar as resonant, seven crossed it as resistant, and six split the room. What follows is what each group has in common.

The one variable that decided everything

People measured quality the way you would measure freshness: by how close the image sits to the tree.

The winning images are a trip to the source. Dew still on the cut fruit. Blossoms still on the branch. Crates, groves, hand-painted signs. The closer a picture sits to the moment the orange left the tree, the more the room trusted the juice inside.

Line all 24 images up by the share of dots they took green, and they sort themselves into a road. One end is the orchard. The other end is the design studio. There is almost nothing in between, and what is there splits the room in half.

Figure 2 · Every image, ranked

What won: the shortest trip from the tree

Eleven images cleared the 60% positive threshold. Read top to bottom they are a drive through citrus country: the cut fruit, the branch, the roadside stand, the picnic in the grove, the crate on the tailgate.

The strongest single result in the study contains no typography, no packaging and no styling beyond a knife cut. That absence is the whole point.

Packaging can win, but only with a warrant

Two packaged images cleared the bar, and they share one staging device. The carton is sunk into a bed of oranges. The bottle is tucked into a crate of clementines. In both cases the real fruit is doing the vouching, and the pack is simply allowed to sit inside it. Put the design on top of the fruit instead and the mechanism inverts, which is exactly what happens further down this page.

One illustration got through

The engraved estate crest is the only drawn image on the winning side, at 77.8%. Its vocabulary points at an orchard: engraved realism, a botanical wreath, an estate mark. It does not look designed, it looks inherited. That distinction turns out to be worth about seventy points, and the next section shows what happens when heritage points somewhere else.

What lost: the orange as an art object

Seven images crossed the 60% negative threshold, two of them unanimously. The more the orange became a piece of design, the less the room trusted the juice.

The first way to miss is to turn the fruit into art. A mosaic orange, a rubber-stamped orange, a papercut hand. Clever images, all of them, and all of them buried. The second way to miss is trend-forward branding. The retro can went thirty for thirty red. To this audience, fashionable packaging design says soda, not fresh-squeezed.

The twin test

The cleanest pair in the data is one orange given two treatments. Same subject, same central composition, similar crowd size. Seventy-six dots landed on the pair, and seventy-five of them agree.

One is an orange. The other is a picture of an idea about an orange. That is the whole study in two images, and the room needed no instruction to tell them apart.

The six images that split the room

Six images landed between 41% and 59%, and every one of them sits exactly halfway between the tree and the design studio. Real fruit under heavy art direction, or hand-drawn labels that are half charming and half distant. When an image is half fruit and half design, the vote comes back half green and half red.

Nothing in the middle is trend-driven. This zone is pure arithmetic: the ratio of fruit to design in the frame, read back as a percentage.

Attention is not endorsement

The disco-ball orange drew 39 dots, the largest single crowd in the study, 14.55% of everything placed on its board. Every one of those dots was red. It stopped people, held them, and was rejected unanimously.

Plot the two measurements against each other and there is no relationship at all. Big crowds sit at the top of the chart and at the bottom of it. This is the number that gets misread most often in the wild, because a stopping-power test and a quality test look identical from the outside and answer opposite questions.

Figure 3 · Crowd size against verdict

What to do with this

The decision logic translates cleanly into art direction, for juice and for any category where quality means fresh.

One warning before you run with this

This study decoded a single attribute, highest quality, which for juice collapses into freshness and provenance. The disco orange and the trendy can might win most fun or coolest brand with exactly the same fifty people. Before you extend any of this into a brand system, confirm which attribute the brief is actually buying.

Methodology and confidence

50 completions across 107 opens, 535 dot placements over 24 images on two boards, 268 on Board 1 and 267 on Board 2, classified at the platform's 60% threshold. Per-image crowds run from 10 to 39 dots. The three anchor verdicts, 36 of 37, 0 of 39 and 0 of 30, are effectively beyond sampling noise, and the distance gradient replicates on both boards independently. Confidence thins in the contested zone, where crowds drop to 10 to 17 dots and scores sit within a few dots of even. Treat those six as directional. The palette observation is correlational: the pink-shifted images were also the most trend-coded, and this study cannot separate the two.

Frequently asked questions